Oral-History:Alex Acero
About Alex Acero
Alejandro (Alex) Acero is an IEEE Fellow (2004) “for contributions to noise robust speech recognition and speech technology education.” He earned his undergraduate degree, Engineer, Telecommunications from the Universidad Politécnica de Madrid (UPM) in 1985; a master’s degree in Electrical Engineering from Rice University in 1987; and a Ph.D. in Electrical and Computer Engineering form Carnegie Mellon University in 1990. He has been granted 160 U.S. Patents, published more than 250 technical papers, and co-authored the textbook, Spoken Language Processing (Prentice Hall, 2001).
He is the recipient of many awards and honors, including a Fullbright Scholarship; the 2017 Norbert Wiener Society Award as well as the 2013 Best Paper Award, and the 2006 Distinguished Lectureship from the IEEE Signal Processing Society. In 2010, he became an ISCA Fellow, 2010 “for his contributions in research, development and education of spoken language technologies.” In addition, he has been an active IEEE volunteer, serving in many capacities, including the President of IEEE Signal Processing Society (2014-2015), a member of the IEEE Technical Activities Board, 2014-2015; a member of the IEEE Board of Directors (2018-2019), and member of the Board of Directors of the IEEE Foundation since 2021.
Currently, Acero leads the speech team in Siri, Apple’s personal assistant for iPhone, iPad, Apple Watch, Apple TV, and Carplay; and he is an Affiliate Professor in the Department of Electrical and Computing Engineering at the University of Washington. Before joining Apple in 2013, he spent twenty years with Microsoft Research, managing teams in Speech, Natural Language Processing, Information Retrieval, Multimedia, Communication and Computer Vision. His team at Microsoft Research built Bing Translator, worked on Xbox Kinect, and pioneered the use of deep learning in large vocabulary speech recognition. From 1991-1993, he managed the speech team for Spain’s Telefónica.
About the Interview
ALEX ACERO: An Interview Conducted by Mary Ann Hellrigel, IEEE History Center, 8-9 June 2023.
Interview #894 for the IEEE History Center, The Institute of Electrical and Electronics Engineers, Inc.
Copyright Statement
This manuscript is being made available for research purposes only. All literary rights in the manuscript, including the right to publish, are reserved to the IEEE History Center. No part of the manuscript may be quoted for publication without the written permission of the Director of IEEE History Center.
Request for permission to quote for publication should be addressed to the IEEE History Center Oral History Program, IEEE History Center, 445 Hoes Lane, Piscataway, NJ 08854 USA or ieee-history@ieee.org. It should include identification of the specific passages to be quoted, anticipated use of the passages, and identification of the user.
It is recommended that this oral history be cited as follows:
Alex Acero, an oral history conducted in 2023 by Mary Ann Hellrigel, IEEE History Center, Piscataway, NJ, USA.
Interview, Part 1
INTERVIEWEE: Alex Acero
INTERVIEWER: Mary Ann Hellrigel
DATE: June 8-9, 2023
PLACE: ICASSP 2023 conference, Rhodes, Greece
Early life and education
Hellrigel:
Today is June 8th, 2023. I am Mary Ann Hellrigel [00:01:00], Institutional historian, Archivist, and Oral History Program manager at the IEEE History Center. I’m in Rhodes, Greece at the ICASSP 2023 meeting. I’m with Dr. Alex Acero, who’s an IEEE fellow class of 2004. He is the [00:01:20] senior director at Apple where you work on the Siri project. I welcome you, sir, and I thank you for helping us document IEEE history as well as mark the 75th anniversary of the IEEE Signal Processing Society.
Acero:
My pleasure.
Hellrigel:
[00:01:40] Thank you. I’d like to begin at the beginning. Where and when were you born? The year is suitable, and we will go from there.
Acero:
I was born in Madrid, Spain in 1961 [00:02:00].
Hellrigel:
That’s a very good year.
Acero:
Oh, yes. Yes. Hey. Good, good harvest.
Hellrigel:
Good harvest. Yes. My year of good harvest.
Acero:
Yes.
Hellrigel:
Your mother’s name?
Acero:
Carmen [00:02:20].
Hellrigel:
What did she do? What was her education?
Acero:
Well, she didn’t go to college; went to high school and then she was—well, that was her education.
Hellrigel:
Okay.
Acero:
Back then, it was not that common. That was also [00:02:40]—growing up in the post-war era in Spain, which was—it was tough, right?
Hellrigel:
Right.
Acero:
Yes. The civil war.
Hellrigel:
Your father’s name?
Acero:
Father’s name is Alejandro. Actually, I have the same name even though I go by Alex. My official name is Alejandro [00:03:00].
Hellrigel:
What did he do? What was his education?
Acero:
He worked in the military, and he went to a military academy. He was an army engineer.
Hellrigel:
Oh, and he [00:03:20] made a career in the military?
Acero:
Yes.
Hellrigel:
When he retired from the military, he just retired?
Acero:
Pretty much.
Hellrigel:
No consulting?
Acero:
Well, he had a business on the side that he was working with his brother, selling appliances.
Hellrigel:
Home appliances?
Acero:
Yes, home appliances [00:03:40].
Hellrigel:
Refrigerators and things like that.
Acero:
Yes. Toasters, dishwashers, all of that.
Hellrigel:
How did he like that business?
Acero:
I think his dad was doing some of that and he got into it. I don’t know. He never told me, oh, that’s my passion, but I just do it.
Hellrigel:
Right. It was also a time in Spain [00:04:00] when a lot of new housing was going up. It was an era with of construction and modernization.
Acero:
Right.
Hellrigel:
So, there’s a market and a demand for domestic appliances.
Acero:
Right. It was a small store. This was before the days of the mega stores, like the Best Buys of the world where you just go there. There wasn’t anything like that anywhere at that time.
Hellrigel:
Right. In my small town in New Jersey [00:04:20] about the time the COVID-19 pandemic hit or just before, our small family-owned appliance store closed. The Lijoi family had a small appliance store in town for more than fifty years. For the most part, the small appliance store and other independent stores are long gone.
Acero:
Right. Well, my dad’s store even before he retired, they just shut it down [00:04:40] because they were already seeing the competition.
Hellrigel:
Yes, correct. The competition from the bigger chain stores.
Acero:
They were older and said, what, I’m done.
Hellrigel:
Did you ever help him in the store?
Acero:
I remember going to the store and the storage unit underneath in the basement. I would not do [00:05:00] anything with the customers, but they would say, hey, you want to help me bring this box here? I said, sure. I just go occasionally and do that, but not very common.
Hellrigel:
Do you have any siblings?
Acero:
I have a brother, two years younger.
Hellrigel:
What is his name?
Acero:
Fernando.
Hellrigel:
Fernando. [00:05:20] Is he an engineer?
Acero:
He’s a civil engineer.
Hellrigel:
Civil engineer. In Spain?
Acero:
In Spain. Most of his career, he’s been doing projects in Spain, but there was a time in which a lot of the infrastructure he was building the highway system [00:05:40] and the road system, high speed train, and all that stuff. They mostly finished a lot of that and then there was a recession. He did a lot of projects elsewhere. He had construction projects in Panama, Colombia, and Saudi Arabia, so there was a time where he was traveling quite a bit.
Hellrigel:
[00:06:00] When you were growing up in Madrid, what was life like for you and your family? What do you remember of your boyhood that you would like to share?
Acero:
Houses in the U.S. style are not as common in the downtown. Downtown is more like Manhattan in the sense that there are apartment buildings. We had the second story [0:06:20] in an eight-story building. On the second floor, we had a condo. We lived there and school was a fifteen-minute walk. [00:06:40] I remember just going to school. School started later than in the US. It started at 9 in the morning.
Hellrigel:
Did school run from 9:00 A.M to 4:00 P.M. or something like that?
Acero:
Well, no. something like 9:00 A.M. to 1:00 P.M. then, we broke for lunch, and we were back at 3:00 P.M. [00:07:00] Then we had school from 3:00 P.M. to 5:00 P.M. or school from 9:00 A.M. to 1:30 P.M. and from 3:30 P.M. to 5:30 P.M. I had lunch at home every day. I just walked back
Hellrigel:
It was a public school.
Acero:
No, that was a religious school.
Hellrigel:
Religious, okay.
Acero:
But at that point, it was funded [00:07:20] partially by the government.
Hellrigel:
Did you have any hobbies or play sports?
Acero:
Yes, I played soccer all my life. I played basketball. I played tennis. We played everything, and skiing. [00:07:40] We liked to do all kinds of sports, bicycling, you name it.
Hellrigel:
Did your family take vacations?
Acero:
Yes. I forget exactly when we did it, but when we were maybe six or five or something, my parents bought a [00:08:00] small home outside Madrid near the mountains. We’d go there for the summer. We spent the summer there and then we would go there for the weekends.
Hellrigel:
Was that common?
Acero:
It was more common in the 1960s because the economy at that point was booming. [00:08:20] In the 1930s, the 1940s, and the 1950s, it was not so good.
Hellrigel:
Sure.
Acero:
The 1960s were a booming time, so probably there was more disposable income. My parents did that and we loved it because it was close to the mountains, so it was not as hot in the summer. There were more opportunities to go around [00:08:40] and be in the countryside.
Hellrigel:
Is this where you learned to ski?
Acero:
No. It was closer to the mountains, but I did not learn to ski there. For some reason, I learned to ski years later, when I was maybe fifteen. At the beginning, we mostly went there [00:09:00] on our bikes with our friends and biked everywhere.
Hellrigel:
Did they close down the shop for the month of August? Europe used to have the one-month common vacation in August.
Acero:
Oh yes. They used to do that, yes. That was the vacation time.
Hellrigel:
When you were growing up [00:09:20], what were your favorite subjects in school?
Acero:
I liked the sciences and math, not surprisingly. I bet that this is not an uncommon answer that you get.
Hellrigel:
That’s quite common. Most of the people I record our histories with liked science and math. Some preferred physics and others preferred chemistry, but most said they liked science and math without listing their favorites.
Acero:
I think I liked all of them [00:09:40]. Maybe the things that were more memorization, such as biology, I didn’t like as much, but I liked everything. I think language, Spanish in this case, was probably my least favorite, but I did well in all the subjects.
Hellrigel:
Did you have to study Latin [00:10:00]?
Acero:
I had to study Latin, yes.
Hellrigel:
Usually, people say Latin was their least favorite subject.
Acero:
Actually, I liked it because I found that there were some rules to it, and learning rules is kind of interesting.
Hellrigel:
And logical. Your least favorite then would’ve been Spanish [00:10:20]?
Acero:
[00:10:20] I think languages.
Hellrigel:
Did writing, grammar, and things like that appeal to you?
Acero:
It’s not like I hated it, but that was my least favorite. Whereas I would go and do the math, and I was very happy. Done, next, let’s go play.
Hellrigel:
Did they give you a lot of homework?
Acero:
I don’t remember a lot of homework. There was always homework, but it’s not like I remembered [00:10:40], oh my God, this was terrible. But, if you’re good at it, especially the sciences, you quickly do the homework and then you go and play with your brother or with your friends.
Hellrigel:
Did you have any hobbies such as taking apart the radio, the TV, or mechanical things?
Acero:
No. We [00:11:00] didn’t have Legos in Spain back then but they had some other brand that was kind of similar. We love—my brother and I would love to build all kinds of things.
Hellrigel:
Yes, they had Lincoln Logs in the U.S. and then—well, in the 1930s, the erector sets.
Acero:
Right. So, [00:11:20] we played with all these things, but one thing that we played forever was something that was kind of Lego-like. Again, [it was a] different brand. It was mostly castles, so we built all kinds of castles. We built the turret and all the bridges and the arches
Hellrigel:
Drawbridges, too?
Acero:
Yes, all these things. We built tons of them. Then we had miniature cars [00:11:40] and trains. Then we would just run and play with that. Remember we were doing a lot of that in the condo. Then, of course, when we went outside, we just kicked the ball.
Hellrigel:
Did you play on any competitive teams?
Acero:
Not on competitive teams. I played a lot of pickup games [00:12:00].
Hellrigel:
Pick up.
Acero:
Yes.
Hellrigel:
When you were growing up what did you think you were going to do?
Acero:
When I was what age?
Hellrigel:
Oh, when you were eight, ten, twelve years old. Did your parents tell you that you were going to go to college? What were their expectations and your aspirations?
Acero:
I don’t even remember that [00:12:20]. Probably, I started thinking about this when I was a little bit older, but I had an uncle who was an engineer. Well, at that point, it was an industrial engineer because there wasn’t really telecommunications or electrical engineer. He was an industrial engineer, but he was doing a lot of [00:12:40] building his own circuit boards. He mounted telescopes. He always would tell me, “Hey, you need to go and be an engineer.” I remember that, but I would just go there and play with him, but maybe I got some of this from my uncle.
Hellrigel:
In school, did you have to take any shop courses [00:13:00]: wood shop, metal shop?
Acero:
Well, we had not quite this, but we had something where you needed to do things with your hands, and it was different projects. Yes.
Hellrigel:
Manual arts?
Acero:
Yes, manual arts. It could be anything from doing something like cutting paper and assembling this to building. [00:13:20] I remember I built a model house of the country house that we had. I built that with cork, cutting it and then gluing in it with some covering that resembled stone or brick that you glue in. and make the roof and all that kind of stuff.
Hellrigel:
That [00:13:40] sounds fun.
Acero:
Yes.
Hellrigel:
Then you’re in high school, so you’re getting a little older and you still like math and science.
Acero:
Yes.
Polytechnic University of Madrid
Hellrigel:
At what age do you decide you’re going to go to college to be an engineer?
Acero:
My uncle and I were thinking because I had a few uncles that were working [00:14:00] with electric circuits at that time. I kind of thought that that was one option. But there was something that intrigued me too, which has been becoming an architect.
Hellrigel:
Oh, and you liked to build castles and structures.
Acero:
Right. Maybe that’s it. For a while, I considered both, and at some point, I can’t remember exactly when, [00:14:20] I said, no, I’m going to be an engineer. And not just an engineer, but an electrical engineer. The school that did that was reasonably new in Madrid. It had not been there for very long, probably eight or ten years, so it’s not like electrical engineering was a well-known field [00:14:39].
Hellrigel:
This is where you’re going to go for your undergraduate education at the Polytechnic University of Madrid.
Acero:
Right, right.
Hellrigel:
Is that the only engineering school around?
Acero:
At the time that I went to school, if you wanted to do electrical engineering, there were just two places in Spain: the Polytechnic University of Madrid and the Polytechnic University [00:15:00] of Barcelona. That was it. If you asked me that question ten years earlier, nothing existed, right. There was industrial engineering, mechanical engineering, and civil engineering, of course, and probably one of the oldest, aeronautical engineering, but not electrical engineering [00:15:20] and computer science didn’t exist either.
Hellrigel:
Why did you select Madrid?
Acero:
That’s where I lived. In the U.S., you hear kids applying to college and they apply everywhere. But back then that was unheard of [00:15:40], particularly if you’re in a big city like Madrid. Now, if you’re from a small town in Spain, well there isn’t a university there. So, then you have to go, and then your choices were probably more limited. If you’re going to do engineering, you have to go to Madrid or Barcelona. If you’re going to do other things, you’re going to study law, or if you’re going to study philosophy or something [00:16:00], then there were other smaller schools that you could go. But if you’re going to study engineering—but in my case, if you’re from Madrid and you want to do engineering, it would not even cross your mind to not go there.
Hellrigel:
You lived at home?
Acero:
Yes.
Hellrigel:
Nowadays, college is so expensive. [00:16:20] Did you get a scholarship or did your folks pay?
Acero:
The price of college in Spain, it’s all public, is much, much cheaper. I don’t even know what the price is now, but I think tuition is probably around $2,000 per year, something ridiculous like that. Back then, it was less [00:16:40] than what you pay to attend a private school. Definitely not a, not a big deal.
Hellrigel:
Yes, less than childcare.
Acero:
Probably much less.
Hellrigel:
And is it very competitive to get in? If it is that affordable, are there limited seats thereby making admissions so highly competitive? [00:17:00]
Acero:
It was highly competitive, but it also depended on the major. There was the university, the Universidad Complutense and then Autonoma and then Polytechnic. There were three universities in Madrid. That was it at that time. You had to take an exam like the SAT [00:17:20]. Then they find a combination of your SAT. It’s not called the SAT, but it’s basically like that. They took that test and your high school grades and took the average. Then a combination of that is that your number. And then they tell you. To enter electrical engineering [00:17:40], you needed to have a number higher than this.
Hellrigel:
Oh, then each discipline would have a number?
Acero:
Each discipline would have a number. At the time that I went, electrical engineering was getting very popular and there were only two places in Spain, so the cutoff grade was probably one of the highest. [00:18:00] If you want to study that, it was quite competitive. Fortunately, I did quite well, so I had no issues, right? But many others couldn’t make it. So, it was quite competitive. But there was no holistic application. It was just these two numbers you average. That’s it. It’s very simple.
Hellrigel:
Did you put together a portfolio? [00:18:20]
Acero:
No, no. It is just a formula. They look at your number and then you’re in. If you want to go there, you’re in. If you have above this, then you can choose anything you want. If you want to go electrical engineering, but you are below then it’s out of the question. Then you have to think mechanical engineer or medicine or something else.
Hellrigel:
What, what was your backup plan if you didn’t get in for EE [00:18:40]?
Acero:
I don’t think I had a backup plan. Well, I knew my high school grades already were quite good, so now I’d need to totally tank, and I didn’t. I didn’t have a Plan B. I suppose my Plan B could have been [00:19:00] picked some other major. But that didn’t really cross my mind. At that point, I knew that that’s what I wanted to do and—Yes. No surprises basically.
Hellrigel:
At the undergraduate level, did you have any professors that were memorable or were mentors? [00:19:20]
Acero:
One of the things back then, there was a three-year degree and then a six-year degree. These were different tracks. You could argue that the six-year degree is more like a bachelor’s [degree] plus a master’s [degree] combined equivalent. At the time, it is not that you said you were doing one [00:19:40] and then you decided you were doing the other. It was a different track from the beginning. I decided to do the six-year degree, and at the end you needed to do the equivalent of a master’s [degree] project. When I did that, I was looking at [00:20:00] what projects [were available]. All the departments there had different [projects and] they had an open house, so you could see the type of demos that they build [and] the gadgets that they do. I saw one that I fell in love with because I found it fascinating. The professors, this group, had [00:20:20] a demo that was building an aid for deaf children. The way it worked is deaf children can speak fine. The problem is in their hearing. They have the vocal capabilities to speak, but because they cannot hear themselves [00:20:40] the speech becomes slurred not because they cannot produce. It’s because of the feedback.
Hellrigel:
Yes. They don’t know how to replicate it.
Acero:
Exactly. They were teaching the children to say, ahhh, and then they played the sound, [00:21:00] but we cannot hear it. They say, ehhhhh. No, not ehh, so ahhh. So, they did an inverse filter of the system to kind of portray what the mouth would look like. Then you see, you get the visual feedback. You’re saying, ahhh [00:21:20], and then you see the equivalent of the mouth and say, no, not this, like that. That visual feedback is what helps them. Now, these days you’d do it with a camera. Right? With a computer vision system. But back then, computer vision was nothing. Right. It didn’t even exist.
Hellrigel:
It just reminds me of the way [00:21:40] some singing lessons are conducted. People would have to watch how to open their throat or whatever. So, you mechanically reproduce it to put the aperture, to put your vocal cords by the way you use your face in the right spot.
Acero:
[00:22:00] Yes.
Hellrigel:
It’s teaching an art.
Acero:
Yes. They were doing this face to face by saying, if you say, ahhh, you’re putting your lips like round. Ahhh. And if you say, ehhh. And, and so—well, you are doing this, and you should do that, or you put a mirror. But now, they’re building [00:22:20] a system so that they don’t need to have the person all the time. And I said, well, how can you do that? And what technology? How is it built? I had no idea how that was built.
Hellrigel:
You’re teaching them to reproduce a sound.
Acero:
Right. And [you are] giving them visual feedback because they don’t get the audible feedback [00:22:40]. They say, well, you should be like this, but you’re doing like this. Oh, okay. So, you changed the sound until you got it. I thought that: A, it had a societal impact and, and B, I had no freaking idea how they built this thing. Well, I was curious. They also had in the same lab [00:23:00] a calculator for the blind where you touch the—it is not one of these small calculators. It was like one of those big calculators that was kind of like a small laptop of those days. You push a button, and you have braille on it. You can touch it and then you hear three times five [00:23:20] and all that was with synthetic speech. Well, how can you push a number and get that it was not recorded? How do you do that? I found it, again, the societal aspect of it, but also the technology. I had no idea, and I was very curious.
Hellrigel:
Wow.
Acero:
So, that’s how I chose my [00:23:40] master’s [degree] project.
Hellrigel:
What was your project?
Acero:
My project was a speech recognition system. Not surprising, that stuck with me for decades. It was an isolated digit recognition system, so unlike today, you can recognize something if you dictate [00:24:00] to Siri and you get all the words. There, all you could say is numbers. You can say 0, 1, 2, and that’s it. That’s all. And up to nine, that’s it. That’s all you could do.
Hellrigel:
This intrigued you and you were curious. You did not know [00:24:20] what kind of lab this would be in. Today, would it be a lab in biomedical engineering or signal processing?
Acero:
Yes.
Hellrigel:
During the talk last night, [at ICASSP, some people spoke about how the signal processing field has broadened and maybe some people no longer identify as “signal processing.”]
Acero:
Yes, yes. This was just the electronics department within the school of electrical engineering. Why there? Because again, that technology was nascent. [00:24:40] It was not even popular there. These professors were using some of this, but what they were teaching, their teaching load, was circuits basically.
Hellrigel:
The three-year track was an electronics degree and not an engineering degree?
Acero:
The three-year track was a technician kind of thing. The six-year track is the joint option. [00:25:00]. Well, not necessarily academic, but it’s a longer program, more advanced.
Hellrigel:
Why did you pick the six-year program?
Acero:
It just probably looks better. You learn more. I don’t know. I got persuaded by oh, you need to do the longer one, the big one [00:25:20], and not the technicians program. Again, I don’t remember much of why I did it.
Hellrigel:
While you are doing this, did you have any part-time jobs in the field?
Acero:
No.
Hellrigel:
No lab work, so you studied.
Acero:
Yes, just study.
Hellrigel:
In the summertime, you—
Acero:
I went to the country home.
Hellrigel:
You went to the country home?
Acero:
Yes, with my brother and my mom, [00:25:40] and my dad would just commute there when he was there. Then when he had the summer break, he spent the whole time there.
Hellrigel:
Did you have any part-time jobs to make pocket money or anything?
Acero:
That was actually very uncommon in Spain. When I talked to my wife, she told me, oh, I was working in the jewelry store [00:26:00] or I was working here in the pharmacy. That was very uncommon. Also, historically, Spain had a very high unemployment rate, so if there was a job let’s bring the person that is unemployed, right? So rare that [00:26:20] teenagers would have jobs. Rare. So never even crossed my mind. My dad would say, well, if you cut the grass, then maybe I’ll give you some allowance. That’s—that was my job. Or, or, or clean up the pool or—
Hellrigel:
Right. And he could have just said no allowance. Just do it.
Acero:
That’s true [00:26:40].
Hellrigel:
You’re doing this and at what point did you decide that you’re going to go to graduate school?
Acero:
This is when I was doing this master’s [degree] project. There was this professor that had a collaboration with a couple of folks. [00:27:00] A professor from the Stanford Research Institute, SRI. They were visiting and they said, oh, Alex, you should meet these two people. Jared Bernstein and Hy Murveit were their names. They had some collaboration and they happened to be visiting that time when I was doing my project and my project advisor [00:27:40]. They’re telling them, “Oh, you guys need to meet Alex.” So, I talked to them and after a while, they said, “Well Alex, you should come to the U.S. to go to graduate school. I said, “Should I?” I had no idea, but they told me that [00:28:00] maybe I should, and they said, “why don’t you apply?” This came mostly through this project. Since I found it fascinating, I said, “Well, let’s do it.” I didn’t know that that’s what I wanted to do, and I applied for a Fulbright Scholarship to come to the U.S. I was thinking maybe [00:28:20] I’ll do Plan B, and I’ll work, just get a job. I hadn’t really thought much about what I was going to do, but because these two professors said, maybe you should do that. I said, okay, maybe I should do it.
Hellrigel:
Now, there are jobs in Madrid or elsewhere in Spain [00:28:40] at this point for you if you got your first degree?
Acero:
Yes. There were lots of jobs because this was a growing field. Electrical engineering, again there were two schools in the whole country. It was new and growing, so that was an area where it would probably be easy to find jobs. [00:29:00] That’s why the cutoff. It was highly solicited for that reason. Well, if you get a degree here, I’m sure you’ll have plenty of jobs.
Hellrigel:
You earned your bachelor’s degree around 1985 or so.
Acero:
1985. I started in 1979 and graduated [00:29:20] after six years in 1985.
Hellrigel:
This degree is a B.E. or a B.S.?
Acero:
They call it B.S. Again, it was not the same. It was like an engineering diploma.
Hellrigel:
Okay, but you are an electrical engineer.
Acero:
Right.
Hellrigel:
You wind up [00:29:40] in Texas.
Acero:
Yes.
Hellrigel:
How did you end up in Texas?
Acero:
I applied for the Fulbright Scholarship. Senator Fulbright started this program.
Hellrigel:
Right.
Acero:
Essentially the idea was let’s [00:30:00] give money to students all over the world so they can come to stay in the U.S. to increase collaboration and reduce the chances of another war because you establish these connections and cultural exchange. [00:30:20] The money that they were giving for whatever reason was decreasing, maybe some politics. The year that I got it, there were only six in all of Spain, but in previous years there were more.
Hellrigel:
Cultural exchange was a peaceful exchange. Wow, you were one of the six Fulbright scholarships awarded in Spain.
Acero:
My year was probably one of the all-time lows. After that, they [00:30:40] got some banks to provide some more money. They administered the program, and instead of the money just coming from the U.S. government through the Fulbright Program, they were adding more from all of these banks.
Hellrigel:
Other revenue, yes.
Acero:
But at the point that I had a Fulbright it was an all-time low. It was very competitive [00:31:00], but I managed to get one of them. Then I said, where am I going to study? I had no idea, at that time there weren’t really rankings that I even knew or even thought about. I just asked some of these people, and [00:31:20], they told me they were in SRI in California. They said, well, Stanford is across the street. That’s a great place. Why don’t you apply there, or Berkeley or MIT? I knew about those schools, and those are the only three that I knew, and only because they told me. Other than that, [00:31:40] the professor that was tutoring me in Spain had done a postdoc in Stanford. Another one that had done a postdoc had gone to MIT. So, those are the only places that I had heard of. I applied, and they asked me the Fulbright, which one do you [00:32:00] want to apply? I said, well, I don’t know. I know about those three. I—said, okay. Then they said, well, we’ll apply to some more. Well, it turns out that none of those three accepted me. I later found out that part of the reason is the grading system in Spain. [00:32:20] It goes from zero to ten.
Hellrigel:
Oh.
Acero:
But the curve is such. I graduated second in a class of 200. I think the way they translated the grades is A was nine and above, [00:32:40] B was – and all that stuff. Even though I got second out of 200, my average was 8.9 or something like this.
Hellrigel:
It meant a B plus in America?
Acero:
Something like that. If you have a B plus, then you’re an okay student, but some of these universities are not going to [00:33:00] take you. Then they figured that out, and I think years later, they changed the mapping system to say, this is not a direct translation you should probably look more at the rank, if anything. That’s probably more informative, right?
Hellrigel:
Right. Yes, more information about where you stand.
Acero:
So, none of them took me [00:33:20]. Then the Fulbright commission told me, we applied to a few more and you got into these other three.
Hellrigel:
Where were your options?
Acero:
One of them was Rice University, another one was the University of Tennessee at Knoxville, and I can’t remember the other one. I had not heard of any one of them [00:33:40]. So, I started looking at some homework and I found out that it seemed that Rice was better than the others and that’s when I found out about these rankings and everything. I said, okay, so I’ll go there and that’s how I chose Rice. Well, it turns out that when I was doing some homework, I said, well, oh, so Rice also, not only is it better in the rankings [00:34:00], but they also happen to have a strong signal processing program. So, from the signal processing point of view, Rice was together with Georgia Tech and MIT. They were probably some of the top universities at that time in signal processing.
Hellrigel:
[00:34:20] So, serendipity.
Acero:
Yes. It was totally a coincidence because that was not planned. That’s for sure. That came up, and I said, oh, this is pretty good. So that’s what I’ve got.
Hellrigel:
What did your family think about this?
Acero:
In Spain, family is a very strong bond, so [00:34:40] on one hand, they said, this may be good for you, but boy, they didn’t like me going.
Hellrigel:
In addition, in their mind you are also going too far.
Acero:
It’s kind of too far, what are you doing? Poor you. They were babying me, but they realized, okay, this is a great opportunity for you. So yes, of course we’ll support you [00:35:00].
Hellrigel:
They’re proud, but they’re nervous.
Acero:
Right. They would rather and probably would have been happier, in a sense, if I had stayed.
Hellrigel:
Could you have stayed at Madrid and got —
Acero:
Yes, and got a job. Yes, I’m sure.
Hellrigel:
Could you have stayed in Madrid and attended graduate school for a Ph.D.?
Acero:
I could have [00:35:20]. I think at that point, a Ph.D. did not cross my mind. It was not very well known and that concept, a Ph.D., at least was not common.
Hellrigel:
At this point, if your brother’s two years younger, he’s already at the University of Madrid in civil engineering.
Acero:
Right.
Hellrigel:
Your parents must have been happy because [00:35:40] their two sons were on their way.
Acero:
Yes. Yes.
Hellrigel:
They must have been happy.
Acero:
Yes.
Hellrigel:
Did they come and visit you at Rice?
Acero:
Absolutely. I spent many, many years in the U.S. and they would visit me every year.
Hellrigel:
[00:36:00] That must have made you feel nice?
Acero:
Oh, yes. Yes. We always had a vacation. They would come and spend some time with me, and then we figured out some trips, like sight-seeing. We did that for many, many years until they were aged, and we just couldn’t really do it. [00:36:20]
Rice University
Hellrigel:
You’re at Rice University then in 1985?
Acero:
Yes. The funny thing is I got the Fulbright Scholarship in 1985, but because of some delay the program actually started in 1986. There was a year when [00:36:40] I finished, but I have not gone [to Rice University]. I spent that year at the university in Madrid. I was teaching a class and doing some research, so in a sense, [I was a] post-graduate. [00:37:00]. It was clear that it was going to be a one-year thing before I went to Rice.
Hellrigel:
When you arrived in Texas, what do you think?
Acero:
It was August. It was hot and humid.
Hellrigel:
And steamy.
Acero:
Hot. [00:37:20]. I remember the first day I was going there with two big suitcases, and I stayed in the graduate house or dorm for the graduate students. Before I saw a vending machine, I was just sweating left and right, and I said, man, I’m going to get, I don’t know, six Coca-Colas [00:37:40], okay. Not just one. First, I got one, and I was going to put my quarter, or whatever that was [in the vending machine] and I said, oh, they have beer. I said, well, I will have a beer.
Hellrigel:
In the vending machine?
Acero:
Then I put in the coins. Hey, that was my first time ever, and I almost spit it out. [00:38:00] I said, what is this? At first, I didn’t care much about the brand. The brand of the beer is “Root,” but who cares?
Hellrigel:
Oh, root beer.
Acero:
That’s how I found out what root beer is, the hard way. It was not beer, but I thought, hey, I don’t know the brands of beer here. That’s the brand, the brand of the beer. Okay. Who cares about the brand [00:38:20]?
Hellrigel:
It’s college. Look, they have beer in the vending machine.
Acero:
They have these things in Spain, so why not in the U.S.?
Hellrigel:
They probably did have them in some places.
Acero:
That’s how I learned. That’s my first experience. This is the first day. Okay. So, root beer is not beer. That’s what I learned that day [00:38:40].
Hellrigel:
Later, if you tried it again, did you like it?
Acero:
Oh, it took me years until I developed a taste for it. Now, I like it, but there was a time that I avoided it, like the plague. I’m not having this. This is terrible.
Hellrigel:
Yes. In America, the old-time sodas are root beer, birch beer.
Acero:
Well, even that, I don’t think I’ve heard [00:39:00] much.
Hellrigel:
Okay. And cream soda.
Acero:
Well, cream soda, yes.
Hellrigel:
But the birch beer was another thing because you’d make it from the roots of the birch tree.
Acero:
That was my first day. I said, welcome to Texas.
Hellrigel:
Welcome to Texas. Well, it was better than if it were some rot gut bourbon or whiskey or some other terrible hard liquor. [00:39:20].
Acero:
Yes. Yes.
Hellrigel:
You’re going to live in graduate housing. Did they assign you a roommate?
Acero:
No, I had my own room.
Hellrigel:
Your own room.
Acero:
And bathroom. It turns out, as I learned later, that was a motel. The university bought it because it was right across the street. [00:39:40]
Hellrigel:
They needed housing.
Acero:
They needed housing. There was a motel. It’s kind of like a motel room. That’s basically what I had. They added a shared kitchen outside, but I had my bedroom and a bathroom all enclosed. That was my housing.
Hellrigel:
And then you’re going to study electrical engineering?
Acero:
[00:40:00] Yes.
Hellrigel:
How is that working for you as you start classes?
Acero:
It went well. I loved it there. It was very, very interesting. I learned a lot.
Hellrigel:
Who’d you work with?
Acero:
I only got the Fulbright Scholarship for one year, like two semesters [00:40:20]. Then I need to figure out how do I finish my master’s in two semesters? I couldn’t really take longer because they didn’t give me more funding. So, I said, well, I’m going to do the course option, but even doing the course option in two semesters is not easy [00:40:40] because I had to take five classes each semester.
Hellrigel:
Wow.
Acero:
I did it, and I finished.
Hellrigel:
By the course option this means you do not have to do a thesis?
Acero:
Yes, not the thesis option.
Hellrigel:
That’s a lot of work.
Acero:
Actually [00:41:00], I don’t have those memories because I think I probably had good fundamentals and it seemed almost easy.
Hellrigel:
Did they have any concern about your English skills?
Acero:
You needed to have the TOEFL (Test of English as a Foreign Language) and pass some bar.
Hellrigel:
So, you [00:41:20] took the TOEFL exam and did well.
Acero:
I took that and must have done well enough that that was above the bar.
Hellrigel:
You had funding for only a year. Did you do anything in addition to studying?
Acero:
No. With five classes every semester I said, well, [00:41:40] I need to finish this. So, five classes one semester and five the next and that’s all I did, just studying. I made friends and was just being another graduate student.
Hellrigel:
Did you like Texas?
Acero:
I liked it. It was really flat. [00:42:00] They somehow, I can’t remember why, but they assigned us host families. I think it’s through the Fulbright connection.
Hellrigel:
Exactly.
Acero:
They probably say, well, in addition to paying for you, we’ll try to hook you up. I remember calling Mrs. Green. She was a stay-at-home [00:42:20] mom and her husband was a doctor. They lived pretty close within walking distance of the campus. She would organize things for us. She had maybe two or three students that she was kind of the host family for. Even though we all [00:42:40] lived in the graduate house, she would take us to her house occasionally to cook or some activities.
Hellrigel:
Maybe barbecue.
Acero:
Yes. Yes, there you go. They also had a little place in Galveston. They took us a few times just to go to the beach [00:43:00]. I remember the first time I had gumbo. She needed to explain what gumbo was because I didn’t know. I said, oh, that’s interesting.
Hellrigel:
Fish soup with sausage.
Acero:
Yes, pretty much.
Hellrigel:
Did you like it?
Acero:
Yes, I liked it.
Hellrigel:
Did you have other unexpected experiences in Texas [00:43:20] besides the root beer vending machine incident. You did not get the beer you anticipated coming out of that machine.
Acero:
The second day, I think I started on a weekend and then they had a welcoming party for all the graduate students. They assigned me a buddy probably an upperclassman in grad [00:43:40] school. I think that they did that probably for everyone. He tells me, okay, come on in. There’s a part to introduce all of them. Do you want a margarita? We’re all above [the age of] twenty-one, so I said, sure. Then he brings me a glass. [00:44:00] I said, where’s the girl? What do you mean the girl? You told me that you’re going to bring me a margarita. Margarita is the name of a girl, a woman’s name in Spain, but in Texas margaritas were a popular drink. I had never heard of that in Spain
Hellrigel:
You thought he was going to bring you a young woman [00:44:20] to talk to?
Acero:
Yes. That’s what I thought. So that root beer was strike one. Margarita was strike two.
Hellrigel:
What did you think when you thought this guy’s going to bring me a lady?
Acero:
Really friendly. Yes. Hey, talk about a buddy to help you out [00:44:40], introduce me to someone one, one of his friends or something.
Hellrigel:
What was his reaction when he realized you thought he was bringing you a young lady?
Acero:
I think he’s probably laughing.
Hellrigel:
Yes. He was like, oh, listen to this new guy. He’s trying to meet young ladies already.
Acero:
Yes. Yes. Then this is more like just learning some of the local language. I had passed the TOEFL, so I probably had good enough English. But some of these terms. [00:45:00]
Hellrigel:
The local terms, the colloquial, and the slang takes getting used to and it is different than the English on the TOEFL.
Acero:
Then of course there’s a lot of Mexican food. The margarita was one thing and then they said, well, so you want a tortilla? Okay. At that point, I was saying, okay, I need to be careful because who knows what’s going to happen, and sure enough, tortilla in Spain means omelet. [00:45:20] So, he brings me this and I said, what is this? “Oh, that’s a tortilla. No, that’s not a tortilla. Oh, yes, it is.” There you go. Strike three. I said, well, I have to do a lot of learning here.
Hellrigel:
You have to watch more of the TV show Dallas [The American television show that ran from 1978 to 1991].
Acero:
Yes, that was the first. Yes, I did [00:45:40] watch a lot of that, but it was dubbed.
Hellrigel:
Okay, so you only heard it in Spanish.
Acero:
Spanish only. Anyway, it’s not just the English, you need to learn all of these terms, right?
Hellrigel:
Yes, you have to know the terms and you have figure out the culture, the slang, and whatever is local.
Acero:
Yes. I took it and said, hey, this is why you’re here. [00:46:00] You’re going to learn a lot. That’s right.
Hellrigel:
Well, at least it was amongst friends and it’s you were not somewhere getting into trouble due to the cultural misunderstandings.
Acero:
Yes. I felt pretty comfortable there and I made friends.
Hellrigel:
Right. Do you recall any professor that you studied with more than the others or was more influential? [00:46:20]
Acero:
There was the department chair, at Rice. His name is Sidney Burrus [IEEE Life Fellow, Charles Sidney Burrus, 1934-2021], and I took two classes with him because. You had to take ten classes, but with him [00:46:40] I took one class my first semester and another class the second semester. He was teaching DSP (digital signal processing). He’s a very well-known figure. I think he’s probably [also on the list of oral history candidates]. I suspect he is on the list, too.
Hellrigel:
Yes, yes. He recorded an oral history as part of the IEEE Signal Processing Society’s fiftieth anniversary history project. [00:47:00]. Now we have been recording additional oral histories to celebrate the IEEE Signal Processing Society’s seventy-fifth anniversary.
Acero:
I took two classes with him and in one of them, he was writing a book. He had the notes, so I [00:47:20] was probably one of the students that had proofread the book. Oh, the draft notes before it was published.
Hellrigel:
It must have been fun that he asked the class to look at the notes. He is well-known in the field.
Acero:
Yes, oh, absolutely. Of course, I’ll do it [00:47:40]. I found a few typos, a few things. Eh, that’s good.
Hellrigel:
Yes.
Acero:
Exciting.
Hellrigel:
Yes. Did you feel welcome [at Rice University]?
Acero:
Yes. I felt like I walked onto the campus, and everyone was smiling. I had a great time at Rice [00:48:00].
Hellrigel:
Did you pick up any of the college culture like going to football games and basketball games?
Acero:
We went to a couple of football games. Rice is a good academic school, but they were not doing well in sports.
Hellrigel:
In sports, not too good.
Acero:
I don’t think they’re in Division I anymore [00:48:20], but they were at that time. They were getting crushed, crushed. It was embarrassing, like sixty to three or something like that. The scores were lopsided. I went to games, but I didn’t even know the rules. You went there and there were two things that were at least worth watching. The band was very good and the cheerleaders [00:48:40]. The football, if you want to watch the other team, not Rice, it was okay.
Hellrigel:
It was painful if you a Rice fan.
Acero:
It was painful. Yes, it was painful. Painful, but it was fun.
Hellrigel:
Did you have the opportunity to explore Galveston.
Acero:
Yes.
Hellrigel:
Then you’re there for one year [00:49:00]?
Acero:
Well, not even that. Two semesters.
Hellrigel:
Two semesters.
Acero:
Yes.
Carnegie Mellon University
Hellrigel:
Then you go back to Spain?
Acero:
Well, so that’s the thing. I was supposed to do that, and my initial intention was I go there [to Rice University], get my master’s [degree], and go back. That was my initial intention. But of course, life is not that simple, right? My [00:49:20]—Sid Burrus tells me, are you sure you don’t want to go to graduate school to get a Ph.D.? At the same time, that professor that I worked with in Spain, his name is Manuel Pardo, [00:49:40] also told me that I should do a Ph.D. He said, you just got the master’s [degree], so do a Ph.D. And I know a couple of people both at MIT and at Carnegie Mellon [University]. Those are the best places to do a Ph.D. in speech recognition. One of them [00:50:00] called me back, the one that ended up being my advisor, Rich Stern (Richard Stern). I must have done well enough that he said, come and stay. You should absolutely get a Ph.D. Apply here. I applied through the standard process, but he basically told me that I was staying, so sometime before [00:50:20] I finished, I can’t remember if it was February or March. I already know, okay, I’m going to do to a Ph.D. But until then, I was thinking I was finishing and going back.
Hellrigel:
Now continuing for your doctorate is an option.
Acero:
I started [at Rice University] in the fall of 1986 and [00:50:40] I finished in May of 1987. I knew in the fall of 1987 that I would be going to Pittsburgh.
Hellrigel:
What do your folks think about this development?
Acero:
They thought that their son was just going for one year. This is not a year, right? This is [00:51:00] at least three to seven years or something. What? So maybe they were not as excited, but yet at the same time, [they are] thinking, okay, it seems like what you want to do, it seems it is probably good for you. Okay. We’ll support you.
Hellrigel:
They didn’t say you’ve already missed one Christmas and one Easter?
Acero:
No, I did not miss Christmas [00:51:20] because I went home for Christmas. I probably did miss Easter, but not Christmas.
Hellrigel:
I understand.
Acero:
I was there for the summer. So, I had the four months in the fall, went back, went home for Christmas, came back, and then after that I went back for the full summer Before I went to CMU (Carnegie Mellon University).
Hellrigel:
[00:51:40] So your mom got the time to visit because you traveled back to Spain. —
Acero:
Yes, that’s right.
Hellrigel:
She knew that you were doing okay.
Acero:
Yes.
Hellrigel:
You weren’t wasting away as she may have worried.
Acero:
That’s right. That’s right.
Hellrigel:
Now, you’re heading north to this industrial city, Pittsburgh, Pennsylvania.
Acero:
Right. We visited [Pittsburgh]. [00:52:00] When I was finished, they came to my graduation at Rice, and we planned a trip. We went to lots of places. I remember that time we not only saw Houston, which is fine, but it’s not necessarily a tourist destination.
Hellrigel:
Yes.
Acero:
We went to San Francisco and LA (Los Angeles), [00:52:20] and New York City. That was like a big trip. Then we ended up in Pittsburgh. Well, my mom insisted, I want to see where you’re going to live and I’m going to go help you find an apartment.
Hellrigel:
Right. She had to make sure that you had linens, clean boxers, and the essentials for an apartment.
Acero:
All that stuff, right, right. That kind of stuff. That’s what moms [00:52:40] do, right?
Hellrigel:
Yes.
Acero:
My parents came and we took that trip. I organized all the places. We did a lot of sightseeing. It was like two or three weeks traveling after my graduation.
Hellrigel:
So did you drive?
Acero:
I drove and my dad drove, too.
Hellrigel:
[00:53:00]. What were you driving?
Acero:
Those weeks I didn’t own a car, so we had a rental car. Eventually I bought a car in Pittsburgh. It was a Maroon Chevy Citation
Hellrigel:
In May 1984, when I returned east from graduate school at the University of California, I also bought a used Chevy, a Chevette. Not my favorite car, but the price was right.
Acero:
It was beat up. I will not say secondhand [00:53:20], maybe thirdhand, I don’t even know. The three weeks we traveled I was just renting cars and just driving from here to there. My dad, for the longest time, was not afraid of driving in the US even though his English [00:53:40] was pretty bad. It’s very bad. But of course, the signs are mostly international.
Hellrigel:
How would you compare your driving experiences in Spain and the U.S.? Did Americans drive too fast or too crazy?
Acero:
No, the opposite. The Spanish drive too crazy, Americans are too slow. My dad was not afraid. The only thing is you still—there are many signs in Europe. You see less. There are more signs [00:54:00] that are international. There are no words. In the US, there are many of them, but there are also some that says, “Do not enter.” Well, if you don’t understand what that means—okay. Now, if they have like the red with a white, then you can ignore the words and you say, well, I know what that means.
Hellrigel:
The stripe, the X out, means no.
Acero:
[00:54:20] But if I say, well, carpool here. What is carpool, right?
Hellrigel:
Yes, the carpool lane. Now, I have to know what carpooling means.
Acero:
That you don’t know. But most of the time, it was with me was more like the days before GPS. So, I was where with the big map. And I was navigating. Now, you do this and now you do that. And [00:54:40] sometimes it’ll be the opposite. He would do it and I drive.
Hellrigel:
In some neighborhoods now they have it. Sometimes there are time restrictions. For example, you cannot make a left between 7 A.M. and 4 P.M. Impatient people know that rule, but you get yelled at or honked at anyway.
Acero:
No right [turn] on red and all that kind of stuff. It was written. Most of the time we were together, but eventually [00:55:00] when they visited me every year, I would go to university, and they just went by themselves, and I had to explain what no right or red meant and all of these other signals that contain words that are not obvious. If you see a sign, well, those signs are international [00:55:20], but some of them with words I had to explain.
Hellrigel:
Did your brother visit with them?
Acero:
Well, not with them, but he visited me also. He did not visit as frequently, but he did visit me also.
Hellrigel:
You’re going to go to Pittsburgh to attend CMU, so what funding were you offered?
Acero:
[00:55:40] For CMU, I applied for another scholarship from the Ministry of Education. Those were not as prestigious, but interestingly enough, they paid more money than the Fulbright and they would last longer [00:56:00]. I got funding for two years—no, actually—no, let me see. That was actually not the first time. The first time my advisor, the first year, he said, well, I’ll give you the funding the first year. Then I got his other scholarship for the last two years and it was more than I [00:56:20] was getting from my stipend from CMU.
Hellrigel:
This is funding from the government of Spain?
Acero:
Right, the funding for the last two years. The first year, my advisor was providing funding. Like at many places, in many places, if they admit you as a Ph.D. student, they’ll give you a stipend and tuition.
Hellrigel:
Did you have to [00:56:40] do any teaching or lab assistant work?
Acero:
I did do teaching assistant work for one of his classes.
Hellrigel:
Which class?
Acero:
I’m trying to remember. I think it was speech communication.
Hellrigel:
One question, when you were a student, [00:57:00] did you have women in your undergraduate and graduate classes?
Acero:
Very few. When I was an undergraduate, we graduated like 200, I think there were like four that graduated.
Hellrigel:
How about at Rice?
Acero:
I think there were a few more. I can’t remember how many, but a few more in our program.
Hellrigel:
[00:57:20]. When you started at Carnegie Mellon how many students were in your program?
Acero:
The thing is that at that time, I don’t even remember even at Rice how many students were there. I just went to my classes, but I think CMU program was bigger than the Rice [00:57:40] graduate program.
Hellrigel:
What did you think of Pittsburgh? You show up, —I don’t know if the Pirates were winning or losing at that time.
Acero:
No, the Pirates were actually in good shape. That was like the best team at the time. I remember Barry Bonds, Bobby Bonilla, and Andy Van Slyke [00:58:00]. That’s when they were quite good. The Steelers had won championships before, but not while I was there, even though there were fans because in Pittsburgh it’s the Steelers.
Hellrigel:
Right. Diehard fans.
Acero:
Diehard. You do well, of course. If you don’t do well, people [00:58:20] still wear their jerseys and everything.
Hellrigel:
Yes. I was living in Cleveland. That’s where I went to graduate school.
Acero:
Oh.
Hellrigel:
There would be diehard fans in all sorts of weather. On game day, Hough Bakeries made dog bone shaped butter cookies.
Acero:
Yes. Normally, Pittsburgh would take the upper hand normally.
Hellrigel:
Yes. It was not the Browns greatest days. While I was in graduate school [00:58:40], that’s when the Browns and Art Modell snuck out in the middle of the night, really snuck out in the middle of the night and went to Baltimore. The truck showed up in the middle of the night and beat the stadium and the offices. You’re at Carnegie Mellon, do you like Pittsburgh and CMU?
Acero:
Yes. I really liked it [00:59:00]. I didn’t know much about it. They would say, yes, they have a bad reputation with the smoke and all the steel.
Hellrigel:
Yes.
Acero:
But they had a mayor, Sophie Masloff, I still remember her name. She had a campaign to clean up the city. A lot of the historical buildings [00:59:20] were stone. You think that it was kind of like a terracotta style color, but actually instead they were black.
Hellrigel:
Yes, from the soot.
Acero:
Yes. When I was there, I don’t know, maybe they cleaned them up with power washing. Not all the buildings, but many of them to the point that [00:59:40] they look in better shape. Also, there was no pollution because a lot of the steel mills shut down. That may not have been by design. It’s more by competition with the Japanese brands. That the steel industry moved there, and it was not competitive. But regardless, it was a clean city. There’s some museum; it was peaceful. I don’t know if you’ve been there, but it’s very hilly. [01:00:00] There was one of the trams, they call them incline. They go through Mount Washington. They had a museum up there where they had pictures of what it looked like in the 1940s. They have pictures at noon, and it was kind of like foggy, [01:00:20] but it’s not really foggy.
Hellrigel:
It is the soot and polluted air, so overcast like the fog in London in the movies.
Acero:
Yes, except that it wasn’t fog. It was smog or whatever. They will say the workers will go with white clean shirts in the morning, but when they came back it was not white. So, you see that, but definitely [01:00:40] Pittsburgh looked nothing like that when I was there. I really liked my time in Pittsburgh. It was really enjoyable. Plus, we lived in my apartment in the neighborhood that was called Shadyside. It is one of the neighborhoods. [01:01:00] CMU is surrounded by Oakland, Shadyside, and Squirrel Hill. CMU was in the middle of one of them, Shadyside, but you go to all of them. It was reasonably dense within walking distance, even though there was a shuttle that would take you, which I would do if it was either too cold or too hot [01:01:20]. Otherwise, I liked to walk because by the time you wait, and you get there, it takes just about as long. Just get some exercise.
Hellrigel:
Who did you work with?
Acero:
I was working with my advisor, Rich Stern.
Hellrigel:
[00:01:40] Three years, that’s pretty quick.
Acero:
Yes, it is pretty quick.
Hellrigel:
How did you do it so quickly?
Acero:
Well, at some point, I had no idea that the first year you’re getting your way around, you’re working in the lab, work in—there was like a speech group, [01:02:00] pretty sizable speech group at CMU, those days and still. And I was working on one problem. And when I was asking, how long have you been here, your Ph.D.? Some colleagues, “Oh, this is my seventh year, six years, eight years, “Like, ah,” so it’s not like that was my plan to do that. But one day, the light bulb [01:02:00] went off, and I said, “Oh, I think, I got it. I think I got it, right.” And you can see light at the end of the tunnel. I think this idea could be—
Hellrigel:
Will be the project.
Acero:
Could be it. And of course, you need to cross the t’s and dot the I’s and do all kinds of experiments. But you can see, well, this is where you’re going uphill, and now you say “Oh, [01:02:40] that’s the finish line.” It’s there. Yes. You still need to pedal, but I think I can get it in. I guess I was lucky.
Hellrigel:
You’re working on the speech project. Three years is quick. While you were finishing your project were [01:03:00] you on the market looking for jobs? What what’s happening?
Acero:
I wasn’t because I was just focusing on finishing. The last few months are the ones that I started thinking, okay, now I know that I am going to graduate. I was thinking I probably would finish my dissertation in September. [01:03:20]. Now that I know that I’m graduating probably in May, or something, is when I thought that I knew it was just right and I started writing. So, maybe it’s time to figure out what to do. I came to the U.S. for one year [01:03:40], and I ended up staying three more years for a total of four. Maybe it’s time to go back. But Rich told me, “Are you sure you don’t want to look for opportunities here? It doesn’t hurt to look.” I didn’t really interview in many places. I think I considered [01:04:00] two places and I decided quickly. The first one was Bell Northern Research in Montreal.
Hellrigel:
Oh, okay.
Acero:
It’s kind of like Bell Labs and it has always been very strong in speech. And New Jersey, well, there was the baby Bell [01:04:20] Labs there on the other side of the border, Our Canadian friends, that was also quite good, I can reckon. And so, they’re hiring, why don’t you go there. So that was at the time that my parents were visiting. I think it’s probably May, I think it was. So well, let’s go there. I’ll do the interview. And we’ll [01:04:40] just do our trip and see Montreal and Quebec, Niagara Falls maybe— Maybe we had gone to Niagara Falls earlier, because that was, I think, I’m sure that we have gone early, they will visit me every year, so we do some trips. But I have never been to Montreal or Quebec, [01:05:00] so we just drove. And I interviewed, the interview went well and they eventually gave me an offer. But something struck me. I was sleeping in the motel. And for whatever reason, that something didn’t quite look right, in the—in the—in the side of the motel. [01:05:20] What is that? And they were like, electric outlets, outside, a lot of them. Said, “Why would you do that?” So, I was so curious that I asked at the lobby I said, “Why, why do you have all these electric outlets, oh so that you can plug in your car?” “What do you mean plug in my car? My car is gasoline.” [01:05:40] “Well, yes, no, no. In the winter, you need to plug it in. Because if you don’t the starter engine won’t start.”
Hellrigel:
Oh, you got to plug into the heater coil that goes into the oil?
Acero:
On the starter, so that— I said, “I have no idea.” That explains it [01:06:00], but I’m not sure, if you need to do that.
Hellrigel:
Yes. It’s too cold. Well, part of it the starter but, but you got the battery. Friends that lived in Nebraska and all that actually be a heater coil that they used to stick down through the dipstick [01:06:20] in the oil tank.
Acero:
Yes, well, I didn’t try to understand too much. So, at the same time, one of the professors at CMU that had actually got his Ph.D., his name is Kai-Fu Lee. [01:06:40] He was a graduate student when I was there, the first year, so he got his thesis in 1988. Then he stayed at CMU [01:07:00] as a faculty member. He was an assistant professor, and he was on my thesis committee. Rich, sure was my advisor, but Kai-Fu was on my thesis committee, and so was José Moura [José M. F. Moura], by the way.
Hellrigel:
[01:07:20] How was it working with José Moura and the other members of your committee?
Acero:
Oh, well, I’ll tell you a story about this. Anyway, I was going there, and Kai-Fu, got an offer from Apple. And he said, “Well, you know, I’m leaving.” He told me, “You want to join me?” Because when I started a group there [01:07:40]. And I visited, and I said, “Yes, you don’t need to plug in your car in California, and so that sounds like a good idea. Looks nice and sunny. So that’s how I went. I ended up going to Apple, instead of going back to Spain which was the default that I had thought that I was going to do.
Hellrigel:
[01:08:00] Did the government expect you to come back to Spain after they paid for your education?
Acero:
Actually, the issue that I had later on was not with a two-year scholarship that I got from the Ministry of Education, it was actually with the Fulbright Scholarship, which is less money for one year. But it turns out that [01:08:20] that part of the agreement for this program was that you do your graduate studies, and after that, you go back to your home country for two years. That was part of, what Senator Fulbright. [When you return home, you bring back to your country what you learned in the U.S.]
Hellrigel:
It is the reciprocity and understanding that you are educated, and you bring it back home [01:08:40] to help others.
Acero:
Right, right. That is the exchange. Now you are an ambassador for the United States.
Hellrigel:
Exactly, the next Kennedy for the - -.
Acero:
Right. The maximum that they let me stay was five years, so I had spent one year at Rice, three years at CMU. So well, I got to Apple [01:09:00], I know I can only do one year, and that was considered practical training from the visa point of view, even though it was a regular job. From Apple’s point of view. They barely knew anything; I got a regular job. But from the visa point of view, my visa was still a J-1 visa with this practical training for one [01:09:20] year.
Hellrigel:
So, that’s why it was only one year?
Acero:
Right. Apple didn’t make contracts for one year. It’s like a permanent job, but that was kind of like a separate thing, visa status. Maybe I should tell you how I met José M. F. Moura
Hellrigel:
Go ahead.
Acero:
[01:09:40] I went back rewinding when I was at CMU and the beginning I was just talking to Rich [Stern]. I needed to get a thesis committee, and once I knew roughly what I was going to do, he suggested Kai-Fu. He also suggested Raj Reddy, who I think [01:10:00] is probably in our archives. Raj had been one of the founders of the Computer Science Department at CMU. He had been one of the first deans and our department chair for the Computer Science Department, and [01:10:20] eventually, the Dean of the School of Computer Science. It’s like one of the originals. He started with speech, even though CMU computer science did many things. This is a case where even though Rich Stern was in the Electrical Engineering Department, a lot of the people in the speech group [01:10:40] were in the Computer Science Department. It’s one of these things where most of the funding came from Raj and his team. Even though, of course, there were people, most of them from CS (computer science), but there were a few that were from EE and one other department, maybe [01:11:00]. Anyway, he told me José and I knocked on the door to introduce myself. I started speaking Spanish, because I thought José Moura. [01:11:20] He looked at me, and I’m speaking. I forgot exactly what I was saying, trying to introduce myself, whatever it is that I was saying, and he doesn’t say a word while I speak. Then I’m waiting for him to take his turn, and [01:11:40] he responds to me in English. He said, “Okay. So, can you say everything from the beginning in English. You are speaking in Spanish, but I don’t speak Spanish.” At that point, I am thinking, okay, let’s see, is there a table I can hide under? This is not a good start.
Hellrigel:
[01:12:00] How about if you spoke Portuguese?
Acero:
Well, it did not cross my mind, because José is pronounced José, but José is a very popular name in Spain. Moura, I think it could be Portuguese, but it could also be Spanish because in the northwest of Spain, Galicia, they have names [01:12:20] like that. So somehow, I should have thought, looks like maybe Spanish, but maybe, it is not Spanish. I didn’t know much Portuguese, so that didn’t even cross my mind. So, okay, this is not a good start. I totally forgot about that story, probably because I was freaking out or panicking. [01:12:40] But José years later reminded me [laughs]. He said, “I was teasing you.” So, that’s how I met José, so there’s not—this is not the first introduction.
Hellrigel:
Did he understand you?
Acero:
He claimed that he understood a lot of what I said. By the way, [01:13:00] I later learned, since I work on languages, that the Italian, Spanish and Spanish in Portuguese are similar, but it’s not a symmetric thing.
Hellrigel:
Right.
Acero:
I can understand Italians better than Italians can understand Spanish. Portuguese can understand [01:13:20] Spanish better than the Spanish can understand Portuguese. I heard it’s because that is the way it evolved from Italy, to Spain, to Portugal, so you understand your ancestors. But you don’t understand it as much—as well. Not that you don’t, but you don’t start as well. In this case, Portuguese people [01:13:40] understand Spanish better. He said, I understood, I don’t know how much but definitely a good amount of what I said. But that’s not the reason I asked you.
Hellrigel:
I studied Italian in high school, and some of the students took it because they spoke Spanish at home, but [01:14:00] kind of broken Spanish. They thought they wouldn’t need to study, but it’s dissimilar enough, especially with the grammar rules. Well, José Moura was on your committee, so you got along.
Acero:
Yes, I made it despite that initial misstep [01:14:20]. That’s an interesting anecdote.
Hellrigel:
Right. You were just trying to be chatty.
Acero:
Basically. I did learn a lesson or two about making assumptions.
Hellrigel:
At that point, José is the big guy [01:14:40] and you’re trying to come in demonstrate that you are not at all shy.
Acero:
You’re right, right.
Hellrigel:
That is a fun story, though.
Acero:
Yes.
Hellrigel:
You earn your doctorate in three years, and again, that’s pretty quick. So, you’re [01:15:00] going to do what next?
Acero:
I remember some people sing in the shower, and I occasionally do, but probably my best ideas have happened in the shower. Instead of singing, I probably would have been five minutes in the shower. Sometimes I’ll be in there, thirty minutes, and I found that [01:15:20] that was probably some of my most productive time, for whatever reason. I’m not sure if it’s the steam or what is it.
Hellrigel:
It’s a place where you can relax and think.
Acero:
Yes, you’re probably relaxing. Your head is just bouncing all over the place. And occasionally that will happen to pour in my dreams. [01:15:40] I remember sleeping, I sleep well. Occasionally, I wake up at three in the morning for no good reason. Except that all I need to do is to get up, pick up a paper pad and I start scribbling. I have to do that, I have to do it now, I cannot wait until tomorrow, it needs to be now. [01:16:00]
Hellrigel:
Do you keep a pad by your bed?
Acero:
I did that for many years. I haven’t done it in a long time, but I did for many years. Later on in life, my wife would tell me, “Oh, I see that you were up at night. I know because I’ve seen all the scribbles, all the algorithms.
Hellrigel:
Well, [01:16:20] not only engineers do that; waking up at night to write down an idea and make notes.
Acero:
Oh, yes. But in my case, it was at least something where I remember that, when I told you this “aha” moment. It was those kinds of things. At that point is like I will be whenever that happened; I will basically sleep three or four hours [01:16:40] for—not that day. But like, of course you don’t solve it in one day.
Hellrigel:
You get wired.
Acero:
You’re wired. You just have to do this. This is on your mind, and your mind is in “bang” overdrive. This is what you’re doing.
Hellrigel:
Yes, it’s why my mind races and I never worked a 9 to 5 job. [01:17:00]
Acero:
Right.
Hellrigel:
Because you can’t turn it off.
Acero:
Right, right.
Hellrigel:
One of my professors [Carl V. Harris at the University of California, Santa Barbara] advised us to keep a little notebook to collect ideas and avoid a pile of notes on scrap paper. If some idea comes to you write it down. That way, you don’t have to worry about remembering it. Then you could try to sleep. [01:17:20]. Your shower thinking experience is fascinating. What is the greatest idea you came up with in the shower?
Acero:
The main idea for my thesis is the problem that I was trying to solve is Kai-Fu system, Kai-Fu Lee, that I mentioned, graduated and that was Essentially, his thesis was a large vocabulary, it was called large at that point. You may fight that assumption today. But at that point, it was called large, and it was continuous speech, in his main contribution was Speaker independent. The systems until then would work [01:18:00] if you’re willing to enroll, meaning speak for like thirty minutes to the system. So, the system will know how your specific voice sounds like when it’s trying to match it. And what he did is his system, anyone can come and here—here’s a microphone go and speak, and it’ll work. It’ll work real time you see the words [01:18:20]. So, there was a demonstration in one of these DARPA workshops, DARPA was funding a lot of the research at that time. And you could see the jaws drop in all the attendees. Wow, incredible, right. My thesis started [01:18:40] because while that was jaw dropping, it only works if you have good high quality close talking microphones. But obviously, this doesn’t seem like something you want to do forever. Would you like to put it there? Like some of these microphones that you put on our desktop.
Hellrigel:
The buttons?
Acero:
Yes, the old ones. You do that and it’d basically stop working. The system that was working so well when you were holding it like this was more like a headset, actually.
Hellrigel:
So, it was the distance?
Acero:
The distance and the noise.
Hellrigel:
Oh, you have to filter out.
Acero:
Right, even though he was not in that noisy place. If it was in a noisy place forget it, okay, nothing will work not [01:19:20] even not. But here it will work. But with a much higher word error rate, like a factor of three, like three times or something like that.
Hellrigel:
You’re going to figure out how to design a better quality microphone.
Acero:
Well, no, the microphone quality is what it is, so is there something you can do to the signal?
Hellrigel:
Oh, to filter out? [01:19:40]
Acero:
To clean it up after the fact. So that was essentially the problem I was trying to solve.
Hellrigel:
Okay, so you strip out the interference?
Acero:
Basically, but in the end, what I decided is to not do that because well, I started by saying, well, you take [01:20:00] the audio, which is noisy, and you do some cleanup. At that time, Rich had told me look at this paper, it was called spectral subtraction. That’s a technique to do audio enhancement that had been proposed a few years earlier. I’ll use it [01:20:20] in my system taking the audio that had some noise and enhance it, so that you could listen to it again. And then we send that audio. That was working to some extent. Then I figured out well, do you really need to do that step because in the end, the recognizer doesn’t care. [01:20:40] You’re going to the frequency domain, you take the audio, you take a Fourier transform, work in the frequency domain, clean up, go back to the time domain. Then again, the first thing that the recognizer did is to take a Fourier transform, it seems like that is unnecessary, doing the word twice. So, it [01:21:20] could save computation, if you just stay in that domain, can we do the operations directly there? That was one little step and said well for that—if we’re going to do that, we can do things that are not spectral subtraction, because spectral subtraction had all these kinds of artifacts, was called Musical Noise [01:21:20]. It is essentially tones, someone’s like to be ‘pi pi pi’ because frequency and all kinds of time was kind of not very good. I said, well, the reason is, because the tones are—each frequency was processed completely independent of each other if you’re close to the noise level, if you’re very high [01:21:40] signal to noise ratio, then it doesn’t really matter. But then you don’t need too much, when you do something is when the signal is close to the noise. The fact that you do the subtraction of these frequencies independently of the rest, meant that there were these musical tones, can we do them not independently of the rest [01:22:00]? Well, you consider all of the frequencies together, jointly, and to do that you need, like some knowledge that the frequency of the signal that is clean at this frequency and that they’re similar, typically, they are not very disjointed.
Hellrigel:
Exactly. Yes.
Acero:
Yes. So, using that knowledge, building a prior model [01:22:20] of what the speech signal looks like in the frequency domain, and use that to do the subtraction of all of them at the same time. That was high level, the idea there, of course, things are more complicated than that but, yes.
Hellrigel:
At this point, the public will see [01:22:40] you got the wheel and the cassette tapes, and then digital. It’s just so invisible, but you’re using mathematics to deconstruct it [01:23:00]. Last night, I listened to some of the ICASSP presentations and some mentioned that the public doesn’t understand what signal processing does because it is mysterious.
Acero:
Right, right. For us, we know, because we build this thing. So, we understand, but most people outside of the field, they think it’s almost magic, as someone as the Sci Fi writer [01:23:20] who say, “Any sufficiently advanced technology is indistinguishable from magic, right?”
Hellrigel:
Yes, I just interviewed Alan Oppenheim who is also a magician?
Acero:
Oh, yes, yes. Yes, he is.
Hellrigel:
He did some tricks downstairs for some of the grad students [01:23:40] who gathered to meet him.
Acero:
He did one for us in the open air fair.
Hellrigel:
Yes, he said he was doing that. And I told him that it sounds like a workshop. He had them hooked. He had a group of about eight mesmerized and they sat there watching and listening. [01:24:00]. It was like Socrates.
Acero:
Yes, yes. Yes, absolutely.
Hellrigel:
So, a year. What did your advisors think; were you thinking so out of the box, that they were just wondering if this was going to work?
Acero:
My advisor, Rich [Stern] [01:24:20], obtained his Ph.D. in MIT, with what he calls perceptually motivated frontends. Essentially, is trying to replicate how the human cochlea in the human auditory system works. [01:24:40] Now, what I did is more like what you say if you’re building airplanes, airplanes don’t flap their wings. In fact, not only don’t they flap their wings, but they also fly faster than birds. So, you could try to build an airplane flapping their wings, and in fact, a lot of the early planes were like that.
Hellrigel:
They tried to mechanically [01:25:00] reproduce bird movement.
Acero:
Right. So, you’re looking at nature for inspiration. Birds can fly, so if I can do something that behaves kind of like birds, I can fly, too. It’s reasonable. The same thing is by looking at dissecting ferrets or mice or other animals. [01:25:20] You can, maybe eventually in some cadavers, see how the auditory system works. It’s not just the ears. It’s the cochlea and all of the different bones and all that kind of stuff.
Hellrigel:
The membrane.
Acero:
The membrane, exactly. So, can we build an electronic version of that? [01:25:40] So, that was because his advisors were, like, in the acoustic era which was like—.
Hellrigel:
The acoustic era was the roots.
Acero:
Yes, like in the 1950s and the 1960s. So, that was what his original thesis was on. Yes, we all have biases, and his biases whether you try something like that? [01:26:00] And somehow, and I can’t remember why I said, well, I will try something more mathematical that is, again, not like how birds fly just a mathematical model that I think would do the job. Today, we call this machine learning.
Hellrigel:
This was another hot topic last night at the ICASSP session.
Acero:
Today, we call it machine learning, but back then, it was not called machine learning.
Hellrigel:
[01:26:20] What was it called?
Acero:
It was called pattern recognition. It’s not that it’s exactly the same in people who debate, but in reality, it was mostly the same all the time. [01:26:40] It’s just the name was not used that way. Even though the name, the term machine learning, was coined by [Arthur] Samuel at IBM in the late 1950s, it was 1958 or something like that. that it was made back then, going that way. It was not used in practice. [01:27:00] It was not used until sometime in the 1990s, late 1990s, it is used big time. So, when I did that, I took a pattern recognition class with another member of my thesis committee, Vijaya Kumar.
Vijaya Kumar taught that class, and I said, “wow, that is fascinating.” It’s essentially what we would say today, is some version of machine learning taught back then. So, I think I want to use these [01:27:40] types of techniques instead. For Rich, I don’t think that was necessarily his cup of tea or he was convinced, but as a good advisor he said, you have this idea, you feel strongly about it, so go for it. Go for it. And, well, it worked [laughter]. [01:28:00] That’s why I graduated [with my doctorate] in three years. Otherwise, I probably would have stayed. By the way, one of the office mates, I was in an office with two other students that had started roughly at about the same time, and one of them was Japanese. His name was Aki Ohshima. [01:28:20] His first name was actually Yoshiaki, and his last name is Ohshima. [01:28:40] So, he was right at the same time, but Aki was doing the perceptually motivated approach. I graduated in three years, and he was there, and we’ll go there. I would visit CMU every year, [01:29:00] sometimes more than once a year, just after I was working and everything.
Hellrigel:
Just to visit.
Acero:
Just to visit. He was still there the first year, and he was still there the second year. I lost track at some point. He was not in the office, so what happened? Finally, he graduated. But not necessarily because he had done a fantastic thesis, [01:29:20] so it was more like, okay, you’ve done your maximum, you get your degree.
Hellrigel:
You earned your doctorate degree.
Acero:
You get your degree, but it may not be a great discovery. In hindsight, I was happy that I chose this more machine-learning approach versus this. [01:29:40] But it’s not like it was obvious when I started that that was going to be the answer.
Hellrigel:
It could have been one going down there.
Acero:
It could have been me. It could have been me. If it was so obvious, of course, he would not have done that, right?
Hellrigel:
Right, he would’ve. Yes.
Acero:
It was not so obvious.
Hellrigel:
Yes. So, you took the risk, you thought [01:30:00].
Acero:
I tried to build the airplane without flapping the wings, and basically, my colleague was trying to flap the wings.
Hellrigel:
Right, well, sometimes—
Acero:
It sounded very reasonable. It’s just that I was more interested in the other approach.
Apple
Hellrigel:
You graduate, and then you’re going to go to Apple for one year. [01:30:21] What are you doing for Apple for one year?
Acero:
I had interviewed there, and I showed up these are the days where everything is printed, of course, there’s no cell phones or anything like that. They sent me this e-mail, I printed it, and I show up. It’s in a different building [01:30:40] than where I had interviewed because, of course, Apple is growing like wildfire, and they relocated. I go there, and I look around. I must have had the wrong address because this is the Cupertino National Bank. So, okay, [01:31:00] I go around the block, and I see only the entrance to the bank. I look around and I keep doing this. There was a stair, no. Then after a while, a big-burly security guy comes out of the bank.
Hellrigel:
Why are you looking around?
Acero:
Well, this is not good. This is my first day. [01:31:20] I haven’t even started, and I’m already in trouble. I didn’t know what he was going to ask me. He tells me, “Are you looking for Apple?” I said yes, I am. He says, “It’s in the second story of the bank.” But I looked there. It’s just there’s tinted glass [01:31:40]. He said, “Well, yes. Knock on the door there.” I knocked on the door, and sure enough, it was on the second story of the Cupertino National Bank, but there were no signs. There was nothing saying Apple.
Hellrigel:
Basically, they rented space from the bank.
Acero:
Yes, but not only that, [01:32:00] it was a secret project.
Hellrigel:
Oh.
Acero:
They could have rented space and put an Apple logo, so they did that on purpose. Who is going to think that we have an Apple facility on the second story of the Cupertino National Bank? No one is going to think that. Right? Especially if we don’t advertise it. [01:32:20] So, it was done on purpose, not, as by design, and it’s all tinted. So, you couldn’t, and there was no way I knew it. But no one else would know either.
Hellrigel:
Why so are they working for DARPA?
Acero:
No, no, no, this was like Apple has always been very secretive [01:32:40]. It was one of these skunkworks projects. Not that the rest were not secretive. At least you could see there’s the Apple logo in front of the building, so this is an Apply building. Now you don’t know what they’re doing in there. But here, it’s like, you, the Apple employees didn’t even know that it existed. [01:33:00] It was that secret.
Hellrigel:
Why were they so paranoid? Were they afraid someone was going to break in at night and steal their ideas?
Acero:
They wanted complete secrecy, even from the employees that they don’t know. This is something that still exists to some extent. They don’t want the average employee to know [01:33:20]. If the average employee doesn’t know, [it is more likely you can keep the project secret.] A lot of the leaks happen from existing employees of all companies, right? They either voluntarily tell the reporter or someone else. Or the reporter is lurking near the buildings, and there is a bar there and the reporter says, [01:33:40] oh, so what is that, Jenny? You’re working on this. And before you know it, they pretend that they work there, too. And before you know it, the secret is out. The best way to keep a secret is if I don’t tell you, then you don’t need to keep the secret because you don’t know.
Hellrigel:
Is this part of Apple culture as well as the other companies in your field? [01:34:00]
Acero:
They are. But certainly, Apple is notorious for being secretive. It is that team, so it’s not like all the teams are like that. But the team that I joined was one of them. And I didn’t know [01:34:20].
Hellrigel:
What did you think then?
Acero:
Well, after I started, my heart rate started going down from the security. [laughter] I said, “Hey, this is pretty cool.” This is like working on a top-secret project. Hey, that must be cool. So, that was the first day. The rest is like, this is a great team essentially [01:34:40] and they were trying to build the next Macintosh. They wanted to keep it totally under wraps because there were other teams that were working on the evolution of the Macintosh. They didn’t want to feel like someone was competing with you. This is like a clean slate [01:35:00].
Hellrigel:
Right. This is maybe a little before that Super Bowl cartoon.
Acero:
Right. The Super Bowl, the big Super Bowl commercial.
Hellrigel:
Was this the big commercial with the sledgehammer smashing it.
Acero:
Well, this was 1984 when the original Macintosh shipped.
Hellrigel:
Right. Silly me, you are working on the next Macintosh.
Acero:
Yes. That was like kind of out of the blue. [01:35:20] I wasn’t there when they did that, but what I heard is that it was something similar because back then, Apple had the Apple One and then the Apple Two. The Apple one is a big deal. But then the Apple Two was more incremental and the Macintosh was like a big disruption. That team was secluded because they were building the Apple Three. [01:35:40] And these guys were working in parallel. And the Mac, Macintosh team knew about the Apple Two, Apple Three all, but not the other way around.
Hellrigel:
Wow.
Acero:
They were trying to do something disruptive, not just incremental. My team was supposed to do the next Macintosh. So, then [01:36:00] Apple One and Apple Two died when the Macintosh— but they were building next versions of the Macintosh. And we’re like, no, we know what the Macintosh team is doing, but they don’t know that we exist.
Hellrigel:
Did you feel a little weird working for a company that was so paranoid or secretive?
Acero:
A little weird, but again, [01:36:20] I remember as a young kid, I played spy also and that felt good. Okay, so, it is almost like a game. You’re playing a spy.
Hellrigel:
You probably had to sign agreements not to talk about things.
Acero:
Oh, yes, yes, yes. [01:36:40] But I didn’t think much of it. Hey, it’s secret, it’s secret. You, they indoctrinate you. These are all the things that you can do, the things that you cannot do all these classes, all that kind of stuff, yes.
Hellrigel:
Wow. So, you couldn’t even tell your family what you were really doing?
Acero:
I could tell them that I was working on speech recognition, which is what I was doing [01:37:00]. But they would not even know, I could not tell them that we’re working on this new version of the Macintosh
Hellrigel:
Right. Because I think the Navy used to have a saying, loose lips sink ships.
Acero:
Basically. That’s right.
Hellrigel:
[01:37:20] You’re going to work there for one year, but it was a permanent job. However, you went to Spain to work for the telephone company to fulfill your obligation.
Acero:
Right.
Telefónica
Hellrigel:
At Telefónica in Spain, you managed the speech team, so what were they doing?
Acero:
[01:37:40] Telefónica was, at that point, the monopoly, the regulated telephone monopoly in Spain, just like AT&T and Bell Labs were for the U.S. before the divesture. [01:38:00] In fact, there was a good collaboration between AT&T Bell Labs and Telefonica, institutional collaboration. When I knew that I was finishing my time well, I’ll go there and I think also, perhaps with my original [01:38:20] connection with Manuel Pardo. And others, oh, own one of the other professors in the department, that is only part-time, works there. Let’s meet him. And they have this job where not only that, but I was managing the team. I had never managed anyone, let alone fourteen people. [01:38:40] But I had a Ph.D. and a year of industry experience in top places and said, well, we want you to be here.
Hellrigel:
Now, could you have told Apple, I have to do this for two years, so could you hold my seat for me [01:39:00]?
Acero:
Interestingly enough, when Kai-Fu hired me, so well, we’ll worry about that later about this visa thing, okay? Somewhere along the way, he told me, let me work with the lawyers here and figure out what is that we can do. There must be something we can do. [01:39:20] And it turns out that there wasn’t much, but there was something. But it needs to come with a petition from some agency in the federal government. I cannot just say I want, or Apple couldn’t say, we want this person. No, no, no.
Hellrigel:
No. There had to be a need by the United States.
Acero:
Some government [01:39:40] official had to do, and that’s the only way out of that clause. Well, it turns out that Kai-Fu had worked on this project when he was at CMU, [and it was] funded by DARPA.
Hellrigel:
Ah.
Acero:
So, he knew a DARPA program officer and said, well, can you do this for me [01:40:00]? And his name was Charles Wayne.
Hellrigel:
Charles Wayne.
Acero:
Wayne said, yes, of course. I knew him a little bit, but Charles and Kai-Fu knew each other very well because, in fact, Charles is the one that funded [01:40:20] Kai-Fu’s project. Kai-Fu’s project, which was called Sphinx, did so well that Charles was very happy, [laughter] and was to some extent indebted to Kai-Fu, said, of course, I’ll do it. So, he tried to say, we’ll, we’ll figure out. Now the government is not [01:40:40] speedy Gonzalez.
Hellrigel:
No.
Acero:
No. Even though he wrote the letter put in the machine, it went from desk to desk. I don’t know what will happen. And along the way, that’s when the Soviet Union was kind of [01:41:00] breaking.
Hellrigel:
Yes, the USSR was teetering and fracturing.
Acero:
Yes. So, I think part of the reason. This is what heard. I wasn’t inside, so I can’t tell. That the Department of Defense this was being run through. The Department of Defense had bigger fish to fry than writing a letter for this guy Alex Acero so that he can stay [01:41:20] in this country, because he’s an imperative for the government or something and Apple is going to help. They were probably busy thinking, okay, what’s going to happen with the Soviet Union? This is going to explode.
Hellrigel:
Instability of that whole region of the world when this break.
Acero:
Well, the whole Berlin Wall. [01:41:40]
Hellrigel:
I watched that on my little black and white TV in my graduate dorm. In 1989, I went back to graduate school for my doctorate. What did you think when the wall came down?
Acero:
It was like, wow. It’s one of these disruptions [01:42:00] that historical moment that you say, well, clearly there’s something to remember. And yes, there is a king here, a king there, a president. But this can be big, right?
Hellrigel:
Right, right. Well, for historians, we, who would have thought it would have happened and would have happened without bloodshed [01:42:20]?
Acero:
Right.
Hellrigel:
Because somebody did not give the command to shoot.
Acero:
Right, right.
Hellrigel:
The only reason why the wall worked was because they had a command to shoot to kill.
Acero:
It happened at that time.
Hellrigel:
Your letter was not going to come?
Acero:
No. Interestingly enough, it finally came. [01:42:40] But it came and I was already working back in Spain. [Laughter] I think I finished in September. That’s, oh, yes, maybe September 1st because if it is five years, September or even August. [01:43:00] Late August, that’s when I started my clock at Rice.
It’s probably five years later, on the dot, in late August, let’s say, that I had to leave, so I started working at Telefónica in September of 1991. [01:43:20] I think I had been there for a couple of months, and Kai-Fu tells me, hey, we got the letter. You can come back. At that point, I could have gone back, but then I thought, well, actually, everything happens for a reason. I’m here now. [01:43:40] It’s an interesting thing. I am back home. I had really not been homesick, but we should give this a try, actually. So, I’ll stay, despite the fact that I could go back
Hellrigel:
Right. You’re back living in Madrid.
Acero:
I was.
Hellrigel:
Back at home?
Acero:
Back at home.
Hellrigel:
Mom is happy?
Acero:
[01:44:00] Mom is happy. I get nice home cooking [laughter] and I don’t need to make my bed [laughs].
Hellrigel:
Laundry service?
Acero:
Yes.
Hellrigel:
She must have been very happy.
Acero:
Yes, yes. After all these years, I finally came home.
Hellrigel:
The prodigal son’s home.
Acero:
Yes, absolutely [01:44:20].
Hellrigel:
You are working on a recognition project for the telephone company but what are they looking to design? Maybe an automated call bank, call center?
Acero:
At that time, they had this strategic partnership with AT&T. So, at that point, you pick up the phone [01:44:40] and it wasn’t press, it was say. The touch-tone had appeared. Press 1 if you want this and press 2 for that. But not all the phones were touch-tones. What if you just say 1 if you want to continue, say 2 if you want sales, say 3 if you want to collect, credit card, and all these things. [01:45:00]. Basically, they want to put these things into operation to remove operators. That’s what AT&T was doing in the U.S. at the time, and in Telefónica we need to do the same, but for our market. So, that was kind of like how they started.
Hellrigel:
Yes. We still deal with that today.
Acero:
Yes, [01:45:20] Yes, you can still see the remnants in some cases.
Hellrigel:
People yelling at their phone.
Acero:
Yes, yes.
Hellrigel:
You’re going to stay in Madrid at the telephone company. How did you like managing fourteen people?
Acero:
It was a shock for me. One of the experiences that I had that made me have thicker skin [01:45:40] is in Spain, contracts are not like in the U.S. [where it] is mostly employment at will, with the exceptions of universities and some government, right? Over there, you have either a permanent position or a temporary one. If it’s a temporary one, it’s a contract for three months or six months or whatever [01:46:00] and at the end of that contract we can renew it if we want. But if you’re a permanent employee, it’s very difficult to get rid of you. [It is] very, very difficult.
Hellrigel:
You don’t need a union because you have a permanent post.
Acero:
The structure, the legal structure, the labor laws. I’m sure they’ve changed, but at least back then, it was [01:46:20] you basically, not to not just punch your boss, but have more than one witness. And like, with much less you’re not coming back.
Hellrigel:
So, the power dynamic is different in management.
Acero:
Right. Very different. Now, I looked at my employees and I liked them. [01:46:40] There was this one that said, well, this guy is just not lifting his weight. He’s not a bad person. But he’s at the bottom, and of course, laying off people there, but except that he was a contract employee. Meaning we can just wait until his contract expires [01:47:00]. So, I go and tell my boss. Can we do that? He said, yes, absolutely. Let’s do that. I agree. We should go for it. Okay. So, we don’t renew it. One day the phone rings. There’s no caller ID, okay, this is like my desk phone. Pick up the phone. “I am “Mr. X”. [01:47:20] Are you Alex Acero?” I said, “Yes.” He said, “You have no soul. You have no heart. You’re just fired my nephew. And you know what? He just got married. They have a mortgage. You put him in the street. [01:47:40] Who are you?” I don’t remember the conversation. I remember, I’m as close to the devil as you can think of why I did that. It’s like, really? I felt horrible. I don’t think I slept for two or three days.
Hellrigel:
[01:48:00] So, he guilted you?
Acero:
I would say at least guilty.
Hellrigel:
But he wasn’t threatening, too?
Acero:
He didn’t threaten me, but he guilted me. I talked to my boss, [who said], “no, you did the right thing. Yes, it’s just not fun, but life is not fun. This is the right decision. Do not revisit it.” [01:48:20] Maybe I had guilt and said, “maybe we can renew the contract, what’s the big deal?” My boss supported me, and said, “no, no, this is absolutely the right thing to do. Let’s do it.” But I didn’t sleep for three days. It was bad.
Hellrigel:
In terms of management, could you have “coached” the person to improve? Currently, “coached” is a management term in the U.S. Do you say you could do X, [01:48:40], you could do Y, and you will improve?
Acero:
I suppose, but I just thought that his intellectual skill was just not there. It was like someone that was just not at the same level as the rest.
Hellrigel:
Right, right.
Acero:
With coaching [maybe he] could improve a little bit [01:49:00].
Hellrigel:
But not a right fit to begin with.
Acero:
But at that point, the easiest thing is to say, the contract ends, just don’t renew it. Let’s find someone else.
Hellrigel:
Right, right.
Acero:
It is just much better, right? And that was easy compared to the rest.
Hellrigel:
You learned pretty quickly that decisions [01:49:20] have consequences.
Acero:
Right. Again, it was very tough. My boss supported me, and said, “You know, this is the right thing to do. Then it was painful, but then I realized a few weeks later my life was better, and the life of my team was better, too.
Hellrigel:
The team could see [01:49:40] their situation would improve.
Acero:
So, well this person is not a good performer. They are not a bad person. They are just not competent, or not as competent.
Hellrigel:
Yes. In the U.S, human resources would say they are not a right fit.
Acero:
Not as competent. Yes. But gone and all of a sudden, now we do not have to pull his weight ourselves. [01:50:00] So they respect my life. I didn’t have to spend time doing that. I spent time, not managing this case, but building the next product or the next research thing. So, after the fact I said, well, that’s a great thing. But it was painful, because it’s the human side. This is the human side
Hellrigel:
[01:50:20] Yes. Hard decisions. Yes, because the uncle implies that you don’t think he is a human being.
Acero:
Right.
Hellrigel:
You have a heart, you’re just a penny-pinching bureaucrat.
Acero:
Right.
Hellrigel:
You are thirty years old and you haven’t had any HR (human resources) training and other things for management.
Acero:
[01:50:40] No, nothing.
Hellrigel:
But you know how a team should work.
Acero:
Right.
Hellrigel:
That’s one thing I was explaining to someone the other day when I was seeing the awards being given for papers three authors, six authors, what, ten authors. In my field, history, most papers and publications have one author and maybe two or three edit a book. [01:51:00] We work more as an individual, but your work’s always been like team.
Acero:
Right.
Hellrigel:
So, maybe management comes with that.
Acero:
But I had never managed, so I think I was doing very well in my game, in the individual algorithms, software, all that thing. But I had [01:51:20] no idea what I was doing from the management point of view. I don’t even know why they brought me in to do that, because I had probably the technical skills, but not the management [experience]. I said I had no experience, and the first experience that I got is like, wow.
Hellrigel:
This team of fourteen, are they all men?
Acero:
[01:51:40] I think there were like two women. They were pretty young, not much younger than me. Maybe two or three years younger because this was a pretty new team.
Hellrigel:
Did you enjoy working for them?
Acero:
I enjoyed it. [01:52:00] I learned quite a bit, but then I also had the reverse culture shock. One day, one of my peers comes to me and he is managing another team. He said, “Alex, so you don’t like me?” I said, “What do you mean good buddy?” He said, “You don’t like me?” I said, [01:52:20] “what are you talking about?” He said, “Well, the other day, our boss came and told me, why don’t you do something just like Alex does?”
Hellrigel:
Oh.
Acero:
He said, “You’re making me look bad. You work too much. You do too much. You do that just to embarrass me [01:52:40] or to make my life harder, right? No.” That was the reverse culture shock. Being Mediterranean, I like having fun, but I work hard, play hard, work hard. I’m doing this, I’m serious, [01:53:00] and maybe I do more than [others], that’s fine. That’s not my problem. I’m not doing this to make him look bad.
Hellrigel:
You were running at a faster pace, and he wondered why you were doing that, and he said you did it to sabotage him?
Acero:
Basically. Then this thing is, maybe there are many good things about being home, being with my family and everything, but the work habits, [01:53:20] maybe it’s not what I’ve seen for a few years. Right?
Hellrigel:
Right.
Acero:
So, is this the right decision after all?
Hellrigel:
[01:53:40] But did you find the American work habits that they worked too much?
Acero:
No because that’s what I’ve done. I felt that it was great, that you focus on this, and you just are not trying to kill time. You’re just going to solve the problem. I thought that was a great idea. [01:54:00] I like it [laughs]. I like the social aspects of Spain more, but I like the work aspects of the U.S., better.
Hellrigel:
And not having really worked in Spain before, you weren’t accustomed to that?
Acero:
Right. I had no idea [01:54:20].
Hellrigel:
You are a manager, you are working on speech recognition, and you had a few bumps. Why leave?
Microsoft Research
Acero:
So, this thing was happening coincidently. Again, life is always a set of coincidences. [01:54:40] Another person at CMU. His name is XD Huang. He writes XD, and sometimes his full name Xuedong Huang [01:55:00]. He got his Ph.D. from the University of Edinburgh, and then he became a postdoc at CMU. He joined CMU in 1989 as a postdoc and stayed as an assistant professor when Kai-Fu left. [01:55:20] When Kai-Fu left, I guess there was an opening, and he took it as an assistant professor and stayed until 1993. Then Microsoft decided to start a research lab. We needed to have speech recognition in the research lab [01:55:40] and they hired him to run it. I had worked with him at CMU, overlapping for about a year in 1989 and 1990. I didn’t know him as well as Kai-Fu, but I knew him. Guess what? He tells me I’m going there, and I need to build a team, so do you want to come?
Hellrigel:
Wow.
Acero:
[01:56:00] At that time, it was also the time when I was feeling the reverse culture shock. So again, something probably happens for a reason. Maybe it is time to consider that.
Hellrigel:
[01:56:20] What is Microsoft at this point? Is it an international company? What did you know about them?
Acero:
I actually didn’t really know much. I thought that they sold MS-DOS. At that point, it had 5,000 people worldwide. It was a very small company. [01:56:40] Apple was considerably bigger when I was there. So, 5,000 people, and they were not known for doing research. Some people said, well, what are you doing when you go there? Are you going to build DOS, which is a sucky thing compared to the pretty Macintosh, and all this. You’re crazy [01:57:00].
Hellrigel:
WordPerfect was the product that I used.
Acero:
Yes. It ran on MS-DOS on the PC.
I visited them. And then of course, things changed when I interviewed because the founder of Microsoft Research [01:57:20] was Rick Rashid. He happened to be a professor at Carnegie Mellon too, working on operating systems. So, somehow Bill Gates convinced him to move [01:57:40]. I said, well, if Rick goes there, he’s not a stupid guy. He’s going there. There must be something because XD goes there, too. Well, from the outside 5,000 people and MS-DOS doesn’t look like much. Those guys are there, [01:58:00] so I need to go and visit. When I went there, I was really impressed with everyone that I talked to
Hellrigel:
And you’re still relatively young.
Acero:
Yes.
Hellrigel:
At this point, Rick Rashid is in his fifties, or is he your age, too [01:58:20]?
Acero:
No, he’s older. He’s probably, I’d say if I were to guess fifteen years older.
Hellrigel:
Gates is buying in a well-known researcher to run his program.
Acero:
Right. I think he was probably—I don’t think he was a full professor. [01:58:40] Maybe he was an associate professor, but he might have been a full professor at that time.
Hellrigel:
He’s bringing Huang in sort of in the middle.
Acero:
Then Rick, basically, just raided CMU.
Hellrigel:
What did CMU think?
Acero:
I don’t think many people were particularly happy. Some people were calling Microsoft Research in Redmond CMU West affectionately.
Hellrigel:
[01:59:00] Wow.
Acero:
Initially, some people called it the CMU mafia, because the first set of people were all CMU. Eventually, they grew to hire people from other places, but they were still the CMU mafia. They were a large concentration. It’s human nature. These are the people that you trust [01:59:20] and they know you, so you can bring them in. When you are building a team like that, no one expects, there’s no research done there. The fact that somebody says well, if this person who I highly respect is there, it can’t be just a fluke. It cannot be just something, [01:59:40] it must be something real. Once you have enough of those, then others say, well, I don’t know, so I’m not from CMU, but I know this guy, and I know, so there must be something.
Hellrigel:
A star led team.
Acero:
This is legit (legitimate).
Hellrigel:
And are they paying an obscene amount of money to get him from CMU?
Acero:
[02:00:00] I’m sure they did.
Hellrigel:
Right. Right. In academia where people aren’t supposed to do things for the money, there must have been some people—
Acero:
Well, but I don’t think he did it just for the money.
Hellrigel:
No, no, no, no. Not just for money.
Acero:
But, but yes.
Hellrigel:
But that’s the reaction with the way.
Acero:
Yes.
Hellrigel:
Like, it’s, it’s a cultural shift. It, it’s like a change from rowing boats [02:00:20] to sail to the age of steam.
Hellrigel:
Someone used the term the other night, a revolution, I forget the right term, the correct term. But it was like a revolution.
Acero:
Disruption.
Hellrigel:
Disruption. A disruption. Right. So, it’s a disruption in the way research scientists are going to work. In the past [02:00:40] sure, you had Bell, you had Lincoln Labs, but now we got—
Acero:
And IBM. Those are kind of the big ones in the U.S.
Hellrigel:
But at this point, they’re waning. So, the research university becomes the real hub, and now we’ve got this guy with deep pockets, maybe.
Acero:
[02:01:00] Well, Bill Gates, everyone knew Bill Gates, maybe the product was small, but Bill Gates, hey, this guy is a little genius. Look what he’s done, right.
Hellrigel:
So, yes, he’s not that young genius college dropout anymore.
Acero:
Yes, right.
Hellrigel:
He dropped out and built a tremendously successful company.
Acero:
Yes, he did. Yes.
Hellrigel:
He’s got a legitimate company and lab.
Acero:
[02:01:20] He still looked pretty young, relatively speaking, I think.
Hellrigel:
Yes. With little glasses and all that.
Acero:
But. Yes.
Hellrigel:
Then you’re going to go out to Redmond, Washington, not to Silicon Valley.
Acero:
Right.
Hellrigel:
And, so, that’s another tension thing.
Acero:
Of course, there was also my current wife in the picture [02:01:40] along all of these things.
Hellrigel:
And where was your current wife?
Acero:
Well, at that point I met her when I was at CMU at one of these parties, a school party.
Hellrigel:
Is it root beer?
Acero:
No, I had gone, maybe it was the real beer at that point. [02:02:00]
Hellrigel:
So, you’re going to meet this lady.
Acero:
No, I just went to a party, with all the students and a common friend introduces us, so, oh, pretty cute. It was in March of 1990. [02:02:20] I graduated in September. I finished. So, I was not there for very long. Right.
Hellrigel:
And you’re off to Spain?
Acero:
No, first I went to Apple, my first stint, right.
Hellrigel:
Yes. Apple.
Acero:
Yes. So, we started dating, [02:02:40] but it’s not like we have been there for a few months. Then I go to California, and she goes to graduate school at Notre Dame.
Hellrigel:
Oh, in the mid-west or at least the middle of the U.S.
Acero:
She was finishing her bachelor’s [degree] at a women’s college called Chatham College [02:03:00].
Hellrigel:
Yes.
Acero:
Then she decided to get a Ph.D. in economics at Notre Dame. That is what she was doing. Yes, we were dating, and we had been together for a few months. Even when we met, it’s not like we started officially dating until two months later. This is a few months, so it was not clear what was going to happen [02:03:20] out of that. Right? We went different ways, but then she visited me when I was in California, and I visited her at Notre Dame. We kept things going, and then I go to Spain. Right. That’s the time where she [02:03:40] was doing the Ph.D. in economics. Well, [she decides] that’s not for me, and then decides to get an MBA instead.
She transferred to the University of Pittsburgh to get an MBA. This is the second year, the first year that I was in Spain [02:04:00] at Telefonica. She was in Pittsburgh getting her MBA. Then she finishes and I’m there in Spain, so that’s when the big step happens. Well, she decides, I’ll go there, follow you, and I’m going to teach English. She was living [02:04:20] with a widow who had like a big condo in Madrid and was renting rooms because her husband passed away. She spends a year in Spain teaching English for some multinational companies [02:04:40]. She had studied Spanish as her minor in college. And of course, she improves her Spanish quite a bit along the way. Then of course, things become more serious, right? Like if someone goes all the way there.
Hellrigel:
What does mama think of this?
Acero:
Well, [02:05:00] it is what it is. This American girl is coming here. Okay. She’s living there and I’m living there. Then this thing comes and of course, maybe there’s an extra incentive say, oh, going back to the U.S, so probably she will say, that sounds good [02:05:20]. And then so when we moved to the—to the U.S. to Redmond, then we moved in together at that point. So, she obviously played a role in, in the decisions, right.
Hellrigel:
Her name? [02:05:40]
Acero:
Donna. Her maiden name, which she never changed her, is Blyshak. If she had changed her name since we spent so much time in Spain, we said, well, what’s your name? Donna Acero [02:06:00]. Oh, are you Alex’s sister? Like, you don’t do that in Spain, right? She kept her name, which is fine.
Hellrigel:
Here last name is Eastern European?
Acero:
Yes. Both her dad’s parents were from the Carpathian Mountain region of [02:06:20] Europe, but the dad came from, what is today, modern day Ukraine out this side of the mountains, whereas the mom’s side came from Slovakia. Now, back then this was all the Austro-Hungarian empire. It was the same, but today is like basically the boundary. [02:06:40].
Hellrigel:
You had fun at the party, you meet your future wife, but you are taking, some time and—
Acero:
Well, I didn’t know at that time she was going to be my future wife. Right. She was just a cute, nice girl.
Hellrigel:
Right. And it’s not like you [02:07:00] ran out after two weeks and got married.
Acero:
Right, right.
Hellrigel:
She’s doing her thing, you’re doing your thing, and all that. Then you wind up in Redmond which I hear is a very nice place.
Acero:
Yes.
Hellrigel:
It’s a fabulous job.
Acero:
Yes. Yes.
Hellrigel:
You stayed there [02:07:20] for a while?
Acero:
Almost twenty years.
Hellrigel:
What were you doing for them, speech recognition?
Acero:
They had a research team, and I was there at Microsoft Research. [02:07:40] Initially, I was an individual contributor, like a researcher, and I was happy that I wasn’t having to deal with the management headaches.
Hellrigel:
[Laughs]
Acero:
Given the experience in Spain. I gave you one story, but there were others, right? Yes, I’m doing research.
Hellrigel:
How is the office or research laboratory experience? [02:08:00].
Acero:
Yes. Well, I think it’s probably just, that’s what the office is like, maybe better or worse, depending on what your office is. But if you want to be the manager, well, you need to get some [02:08:20] battle scars and thick skin and learn not just the technology, but how to manage people, which I didn’t know. Hey, I am very happy that I didn’t need to do that.
Hellrigel:
You’re a researcher who’s recruited by Huang.
Acero:
Right, Huang. [02:08:40]
Hellrigel:
And you feel pretty good that you’re being recruited for this team. They are well known, and Redmond is a nice place to end up. [02:09:00] You’re a researcher at first. Then twenty years is a significant amount of time.
Acero:
Right.
Hellrigel:
And are they as quiet and secretive about projects as Apple?
Acero:
No, no. Quite the opposite. [02:09:20] It was essentially a university department that was run inside a company. Rick Rashid, since he was a professor at CMU, brought the reasonable person principle. It was something that I hadn’t heard of. [02:09:40] And it’s essentially something that is cultural as part of the computer science department at CMU. And you only know what it is once you’re there. Essentially what it is, is we’re going to trust the heck out of you because if you’re reasonable, you can do anything you want [02:10:00]. As long as you’re reasonable. If you stop being reasonable, then we’ll take the, your freedom.
Hellrigel:
If you’re a prima donna or—
Acero:
If you just don’t do whatever reasonable whatsoever, these are what reasonable things are. These are what things are not reasonable. I give you a few examples, and now I empower you to do whatever you want and say, well, hey, they trust me. [02:10:20] And yes, of course, now you realize that you need to think about what you’re doing. If you don’t think it’s right, because you are the one, it’s not like before, it’s like, someone not my problem, someone else will make the call. And that was something that he did. So, he built the computer science, the Microsoft Research team [02:10:40] founded with the principles of the computer science department at CMU. So, it was great. Also, the CS department at CMU was very collegial. It’s, well, it’s done great over the years, obviously. And you are interacting with very smart people. [02:11:00] It was really fun. So why would you consider not doing that? Sounds great.
Hellrigel:
What’s your work life balance? Do you work eighty hours per week? What’s going on?
Acero:
I’ve always worked a lot of hours. I don’t know if it was eighty, but historically, like 8:00 A.M. to 7:00 P.M. [02:11:20] would be pretty common. Lunch would be like, yes, you’re not talking about soccer or NBA or something.
Hellrigel:
You are at your desk.
Acero:
You’re talking and working. Then sometimes on weekends you do something at home. I worked a lot. Occasionally, I would get a phone call from Donna, and she said [02:11:40] honey, you’re planning to come home for dinner, are you not? Oh my God. I lost track of time. Right. This would not happen once or twice. It happened multiple times, despite the fact that I normally watch, but sometimes I like just so much in my thing is like, and before I realized, really? Oh my God, I feel terrible [02:12:00]. The following day, of course, I need to buy some flowers or something [laughs]. She understood. Hey, I know, that’s fine. She was working too.
Hellrigel:
So, what was she doing?
Acero:
She got an MBA, so she was working as a business project manager for a medical device company [02:12:20] that was called MediSystems. It was based in downtown Seattle. Her work was a more fixed 8 to 5 type job, except that she commuted on the bus because in downtown traffic, it’s not only the money, [02:12:40] it’ll take you less time to take the bus. She’ll drive to the park and ride and take the bus. Since she was working and had to leave early, we left at the same time. The techies oftentimes doesn’t start until later. But she’s leaving, I’m leaving at the same time, not going to stay sleeping, right? Right.
Hellrigel:
[02:13:00] Sure, if you stayed in bed sleeping that might really get you in trouble. [laughing]
Acero:
Yes. Yes. So, no, just do it. We both get up at the same time. We leave at the same time. Her start time is more fixed and mine is flexible. But whenever I get up, we both leave at the same time.
Hellrigel:
At what point do you get married?
Acero:
Actually, we didn’t get married until 1997 [02:13:20]. I proposed and we got engaged in 1995. We had been living together for two years and change and I got engaged in November 1995. [02:13:40]
Hellrigel:
And she knows when she marries you that you have this work schedule.
Acero:
She knows that, and it is fine. That’s what you do. You like it.
Hellrigel:
As you mentioned, it is what it is.
Acero:
She realizes that I like it. So, it’s not like, if someone is like whipping [02:14:00], say, “Hey, do this.” Say, well, Alex, you should not. You like it, you have fun, you’re excited, that’s your career. That’s fine. Of course, if she’s cooking dinner, I appreciate it. And we never had an early dinner. It’s not like we had 5 o’clock dinner. I remember 7:30 [02:14:20] is probably the earliest we have dinner. 8 o’clock was more common. Many times, after that, she had lived in Spain for a year. And she liked that style, so not a problem that as long as I didn’t come much after eight, she was okay.
Hellrigel:
What are you doing for [02:14:40] for fun? Hobbies or anything?
Acero:
Well, I played soccer. I was in a team.
Hellrigel:
A Microsoft team?
Acero:
No, there was some other team. I can’t remember how I got into that team, but one or two teams, I was doing that.
Hellrigel:
[02:15:00] And, you had vacation time?
Acero:
They gave me vacation. It was three weeks in the beginning. I think you accrue more, and eventually it became four.
Hellrigel:
Right. You are in the lab, but do you have to travel a lot to collaborate with different teams?
Acero:
[02:15:20] Well, initially Microsoft Research was just based in Redmond. Later they opened labs in Beijing, [China], Cambridge, UK, and then in Bangalore, [India], and in Boston. But when I was there, and for many years, it was just the Redmond lab. We travelled just like professors to [02:15:40] go to conferences and to visit universities and give talks. I travelled as much as a professor, maybe not as much, but quite a bit. I travelled to conferences like ICASSP, Interspeech, SLT, ASRU or some other conference. Then some professor invited [02:16:00] me to give a talk at the university, so I’d go there. So that kind of stuff. That’s the only travel that I would do.
Hellrigel:
At this point are you having any second thoughts that you should have gone into university teaching instead of industry?
Acero:
No, that was the best because in this role professors have to do three things. [02:16:20] They have to teach, they have to do research, and they have to get money. I’m not going to say how much time, but let’s say one third each. So, I didn’t need to teach, and I didn’t need to spend time raising money. I spent the whole time doing research. And I— [02:16:40] the only thing I need is to tell my boss what you think? It’s a good idea? So, getting the funding is a pretty quick thing, basically, your boss says, yes, that sounds good. Go.
Hellrigel:
Okay. What projects are you working on?
Acero:
I work on research that many professors at CMU [02:17:00] and others would be doing, something similar. But at the same time, we’re trying to also see if we can put some of the technology into Windows. So we build—we have a couple of software developers in the team to help us take the algorithms and put them inside an API. And you could do speech recognition [02:17:20] on a PC back in 1995, Windows 95 came out, guess what? You can do this also.
Hellrigel:
Right. So, you are getting funding. Did you start to become a leader of a research team? How does your career progress?
Acero:
So, it didn’t—it didn’t cross my mind. I should be in academia. [02:17:40] Hey, why should I even do that? That was—this is great. I like doing research.
I was a researcher. Then I became a senior researcher [02:18:00] and then a principal researcher. At that time, in the year 2000, the company decided that they wanted to start a product team doing speech.
Hellrigel:
Oh.
Acero:
Until then, it was just research. I said, well, this is time to do this. They asked Xuedong. I said, “Well, hey, why don’t you lead it?” [02:18:20] So then basically he needs to find a person for his former role, and he tells me, “Hey, you want to do this?” What was I doing until now?
Hellrigel:
So, you’re going to move up?
Acero:
Yes. And then I start managing the team in 2000 [02:18:40]. At that point, I had learned more, and I said, well, maybe now I’m ready to manage. I wasn’t ready right after getting a Ph.D., but now I’m ready. I think I learned enough.
Hellrigel:
You’re ten years from earning your doctorate.
Acero:
Yes.
Hellrigel:
And you’re still content and [02:19:00] this is like a project product team. So now they’re talking about more commercialization.
Acero:
Right.
Hellrigel:
At this point, how many people are you managing?
Acero:
I think it’s probably about a dozen or so.
Hellrigel:
Okay. There are different teams working on different projects, [02:19:20] so how many scientist, engineers, and other researchers work at Microsoft?
Acero:
When I started Microsoft Research was very small. There were fifty people; I think they hired quickly. By the time I was managing the team, there was a lab in Beijing that had opened [02:19:40]. So, it’s probably like 200 or 300 people by then. I would say it was a mini Bell Labs.
Hellrigel:
Wow.
Acero:
Again, Bell Labs was at that point no longer in their prime.
Hellrigel:
Right. Bell Labs had been reduced.
Acero:
Right. Many people had left. This was the best basic research lab [02:20:00] in the computer industry, research lab. IBM was good too, but no one, at that point, for a number of years, that was the place to be. But it wasn’t obvious when I first went there.
Hellrigel:
No, no. So, you take a risk, you go there and ten years in, okay, you’re managing [02:20:20] it and such, and eventually though, you’re going to start to look elsewhere or?
Acero:
No, I didn’t. But it turns out that, again, serendipity well, Siri launches in 2011. And the way that works is, [02:20:40] Apple bought a startup whose name was Siri, which in itself, had been a, a spinoff of SRI, and Apple buys it, and it was a twenty people startup. And essentially it was an app. [02:21:00] So you go install the app, launch the app, and now you, there’s a big microphone button. You push a microphone, and you speak and things happen. And what Apple did in the first year after they bought it in 2010, it’s like, well, we’re going to do that, but without you having to go scroll open the app, wait until it launches, and then push a button doesn’t save you any time. By then, you’ve waited—you’ve [02:21:20] spent 15 seconds, 20 seconds, right? No, when you push the big button at the bottom, you launch Siri immediately. That means that now it’s integrated into the operating system. So that’s what they did.
Hellrigel:
So, then it’ll be voice from the get go.
Acero:
Right. You didn’t need to swipe. The idea is voice is an accelerator, [02:21:40] but if it takes a while until you open it, it’s not accelerating. So that’s what they did in 2011. It turns out that the speech recognition that was powering this was done by Nuance, a company that at that point was both in Boston and the Bay Area [02:22:00]. There was Nuance and ScanSoft they merged it was powering. You essentially take the audio from the iPhone, send it to a server in the cloud or data center, and then the recognition and everything comes back. So, that was Siri. [02:22:20]
Apple realizes, well, this speech recognition doesn’t work that well. We cannot rely [on an outside contractor], we think this is core. Apple has always thought if something they think is core, we need to have it in house. If it is not something that is core, and it’s a more a commodity, we’ll contract. [02:22:40] We buy the glass from this guy; we build this capacitor from that guy. We don’t think its core, so we’ll just buy it from the rest. But if this is core then we need to do internally. So, they asked me to go and build that team. I was happy at Microsoft, but Apple of course had a big place in my heart [02:23:00] because I spent a year there. I had a great time, and I liked the culture of the company, everything. So, I said, “Hmm, maybe this is a good time for me to consider it, right?” Also, it would be like doing research. Well, I love doing research, I still do. But I’ve done it for many years [02:23:20], twenty years. And this is an opportunity for the longest time to make a difference in products, right?
I tell this anecdote that somewhere around maybe 2005 we’re having a neighborhood party and one of the neighbors, a lawyer, and the one a doctor are talking there in the summer [02:23:40] with a glass of wine and just chatting. So, Alex, tell me again what you do. I tell him, well, speech recognition. Speech what? Speech recognition. What’s that? I explained to him, and he understands half of what I’m saying, even though I’m trying to dumb it down. [02:24:00] Then he says, oh, you mean like R2-D2 that talks back, that speaks like this robot in the Star Wars movies. So no, that’s speech synthesis. Speech recognition is kind of like your ears, speech synthesis, kind of your lips. Oh, I get it. I think I get it. The other day we lost power [02:24:20]. Seattle loses power frequently in the wintertime when there’s so much rain. The trees, it’s the—
Hellrigel:
Oh, they fall.
Acero:
They saturated, then the wind comes and guess what, they fall on the power line. Right? It’s pretty common. The other day we had rain and a fallen tree, so I called the utility company to report it, [02:24:40] and I talked to a stupid machine that didn’t understand me. Is that what you do? So, you want more red or white? [Laughs] I am trying to change the subject. The answer is that’s true. The technology was there for demonstration, but it didn’t really work. So, I was working on research, [02:25:00] because there were a few companies like Nuance, but by and large, it wasn’t really mainstream by any stretch of imagination. But when Siri came out, I said, well, this may be the opportunity. Maybe now things are ready. I always was very interested in the practical applications. That’s how I even got into the field, [02:25:20] the deaf children and the blind. I wanted to have this technology. I was curious about the technology. I had the curiosity of a researcher, but I also wanted to make a difference. The technology was just not there. They had the prototypes. There were some cases where it could be used, but not mainstream. It was just very, very niche, but [02:25:40] now with Siri it’s Alex, what do you do, speech recognition. Oh, you mean Siri? I said, yes. I didn’t need to explain it. It was kind of the beginning of being mainstream. I said, well, how many places can do something like this? [02:26:00] This may be the time for me to do it. I need to move to product, and within that one year I was in a product group.
Hellrigel:
Right. So, you already made that transition.
Acero:
It’s not like it’s a jump or a leap in the vacuum completely, right? I’ve done this, so maybe [02:26:20] I should take this maybe once in a lifetime opportunity because it also turns out that at the same time, I have been fortunate to be part of a research revolution in deep learning. Now you talk about deep learning, machine learning. Neural networks have been developed for a number of years [02:26:40], but they were kind of dormant. In the 1980s and 1990s, there were some people playing with neural networks. In fact, that officemate of mine was doing not quite neural, but something related. Clearly, the ideas were great, but there was not enough data, not enough to compute. I didn’t know that at the time. [02:27:00] I thought it was just a waste of time. It was just a dead end, right. But then Geoffrey Hinton, one of the grandfathers of deep learning, published this paper in 2006. One of the researchers in my team, Li Deng, tells me, “Hmm, [02:27:20] this is something that may be big.” He tells me I’d like to invite Hinton to come and visit us so that we can learn more about it. Okay, what are you planning to do? Well, Hinton has published this research on this corpus called MNIST, which is the handwritten digit recognition. [02:27:40] You can use this deep learning, which is neural network to do it so well. Okay, so let’s see if I get it right. You are in the speech team. You want to invite Hinton to help us on digit recognition, which is images. But you’re in the speech team, is that what you want me to do? Yes.
Hellrigel:
[Laughs].
Acero:
Of course, something crossed [02:28:00] my mind and I thought this doesn’t make any sense. I’m the speech team, this image, why are you doing this? But then I said, wait a second. I remember when Rich [Stern], my advisor was telling me, maybe you should do it this particular way, and I suggested this other way, [02:28:20] and he supported me.
Hellrigel:
Wow.
Acero:
Maybe I should do the same thing. I actually also then remembered, Raj Reddy had done something similar with Kai-Fu Lee. Raj had more like what we call the old-style rule-based AI, and Kai-Fu came with more statistical models. [02:28:40] Raj was kind of like telling him, I don’t know if that’s going to work, but you feel so passionate about it, I’ll support and help you as much as I can. So, I had seen those two cases. I said, well, this may be a total waste of time.
Hellrigel:
But you’re willing to try?
Acero:
But I’m going to try. [02:29:00] Not only that, but I was also probably thinking that this was going to be a waste of time because I had seen it. It’s not like I hadn’t seen neural networks fail before. I have. If I hadn’t, I might be, but I thought this was going to be a waste of time. But what, it’s not like we’re wasting ten years of ten people. It’s just a couple of months. [02:29:20] Fine, let’s do it.
Hellrigel:
Does Microsoft think you’re jumping to the enemy when you go to Apple?
Acero:
Absolutely.
Hellrigel:
Did some people stop talking to you?
Acero:
Yes. Let me finish this first. I finished it, and we published it. First of all, we got the results. It’s like, wow [02:29:40], 30 percent reduction in this database. That can’t be true. That’s too good to be true. You probably made a mistake. Maybe you tested on the training data or something. That can’t be true. I asked another, whether you go and look it up, and yes, it’s true. I said, we’re onto something. The team put more people working on it, [02:30:00] and it kept getting better. One of the things that you can see is the term deep neural networks. Now you hear it very commonly, but it wasn’t really popular. We published this paper in 2010, 2011, and the whole field of speech recognition moved to these neural networks very, [02:30:20] very quickly. If you were to type deep neural networks in conferences like ICASSP, before then there wasn’t actually zero because that term machine learning had been coined earlier, but no one was really using it very much. But after that one, Boom, so this was a disruption. [02:30:40] Not only Siri is a disruption, but potentially it now enters the popular lingo. People know what it is, they start to use it. This new recognition technology may also make also a big change. So that makes it accurate, a perfect storm, this is the time to go and do it. Now, if I wanted to do the product, [02:31:00] I could have stayed at Microsoft, too. It was going to be mobile. This was going to be on a mobile phone. It’s not on the Windows desktop. We have tried that. The keyboard is too good of an alternative for people to pay much attention. Right. But the phone so small.
Hellrigel:
[02:31:20] Well, the phone, and then this is when people’s cars start talking to them?
Acero:
Well, that was probably a little bit, there was some of that, but by this point it was mostly the phone and the iPhone took over. It was like another big disruption. I could have stayed at Microsoft, and they had a Windows phone called Windows Phone, [02:31:40] but it was like too little, too late. There’s not going to be room for that many people. They called me here and the iPhone is the pioneer, I’m just going there. But I took a big risk.
You tell me, friends. Well, so I told my boss that I [02:32:00] got an offer for this, and it’s not like I was saying it to negotiate. It’s like I was trying to be nice because I really liked my boss then. By the way, I had changed bosses at that point. A few years later, I didn’t report to XD Huang, my boss was Rico Malvar. [02:32:20] He might also be on your list, if not, Rico Malvar; his full first name is Henrique. He is Brazilian and another big contributor here in signal processing. I like both XD [02:32:40] and Rico, great people. I owe a lot to Microsoft, so I felt like I wasn’t playing games. I was just telling them I got this offer. I’m conflicted.
Hellrigel:
But you’re not negotiating to have them make a counteroffer. You made your decision.
Acero:
Not negotiating. On one hand, I didn’t want to leave, but at the same time, I said, well, there’s not that many opportunities like this. [02:33:00] Technology’s improving, and finally becoming useful, or at least you could see it could become useful. How many of these chances do you see? Not that many, right?
Hellrigel:
And they’re asking you.
Acero:
And they’re asking me, and I know the company, I’ve been there before, so I know. So, I jumped.
Hellrigel:
At this point, you’re around forty years old.
Acero:
Yes. [02:33:20] I tell this and before I know it, he said, well, you are going to go to Steve Ballmer’s office tomorrow. At that point, Bill Gates had stepped down and Steve Ballmer was the CEO. And with Bill Gates, I had been in the same room and talked to him in person. We had many reviews, [02:33:40] but Steve Ballmer was more like a businessperson. So, I had seen him in company meetings and at big events, but not at a one-on-one meeting ever. My first one-on-one meeting with Steve Ballmer was when I basically tell him that I’m leaving or I’m seriously thinking of leaving. I had heard stories that he would get mad [02:34:00] and throw chairs. In fact, Kai-Fu Lee was with Apple. Apple went through tough times in the 1990s, so he left and ended up at Microsoft, interesting enough. He was heading the Beijing lab in 1998, when we started. He had been at Microsoft, and he left to go to Google. We hear hard words about Steve Ballmer; that he’s hard.
Hellrigel:
Hard words for—
Acero:
You know, [he has a] temper, and [02:34:20] we hear about throwing chairs. It’s unclear whether he (Ballmer) did that or not, I don’t know. When I was there, Ballmer was very nice to me. He was trying to understand why I was considering leaving, and what he could do to persuade me to not do that, so it was hard. [02:34:40] I said, oh, this guy is nice. He said, well, think about it. I’ll do this. And, and of course he said, well, I’m going to increase [your salary], take up my checkbook and see if that encouraged me to stay. Right? And we’ll make a promotion. We’ll do all this and so, but—
Hellrigel:
Did this encourage you to stay or at least reconsider? [02:35:00] You left, so was your mind made up before the meeting? He was saying, we can make it work.
Acero:
Yes. We can make it work. So of course, when Apple offered me [the job] there was an increase. but at that point, it was not clear that it was [02:35:20] of course with stock. It depends if one stock does better, it could better. But it’s not obvious that one offer is better than the other. So, at that point, it’s not about the money, and it’s not that it was originally about the money. It’s more about this opportunity; a once in a lifetime opportunity. Maybe there are not that many, so you want to do this. [02:35:40] I said, I’m still going, so he brings me back one more time to see if he can convince me. I said, well, I decided, and the following day there’s a lawyer. He’s sweating up the stairs running, and I look at this big folder of documents, of papers. [02:36:00] Essentially, it’s not quite a lawsuit. I don’t know, legalese, but it's kind of like a pre-lawsuit from what I could tell.
Hellrigel:
Against you.
Acero:
Against me because I’m breaching my non-compete agreement.
Hellrigel:
Oh, my gosh.
Acero:
Now this was not a shock to me. I knew it could happen [02:36:20] because I had a lawyer, and this is exactly what happened to Kai-Fu when he left. Whether he (Ballmer) was throwing a chair or not, I don’t know, but he got a lawsuit, preventing him from starting to work at Google. [My lawyer said] the same thing could happen to you, so you need to be prepared. I was prepared. I said, well, it’s not like it’s happening for sure, [02:36:40] but it could. There is a 50-50 chance, and sure enough, it did happen. In theory, my non-compete, which when I came from Telefónica in Spain, I signed, and it was several pages. I just signed it, and I didn’t probably read. And if I read it, I didn’t understand it.
Hellrigel:
There we no ramifications with the non-compete clause with Telefónica. [02:37:00]
Acero:
But twenty years later, well, this is what you signed, do you see?
Hellrigel:
You had to agree that you would not work against the company by working for a competitor.
Acero:
In twelve months.
Hellrigel:
You could not work for a competitor for twelve months?
Acero:
Now I already knew that these were the rules. While talking to my lawyer, I learned that typically, even though that’s a contract, [02:37:20] you often settle for less. It depends on the case, right? So, I had to be prepared in case that happened.
Hellrigel:
Maybe at this point are they still trying to keep you from leaving?
Acero:
Well, I think it’s more like if you are going to leave, [02:37:40] we want to avoid that, but if you do, then we want to slow you down.
Hellrigel:
Their point is, we’re going to make it hurt a little bit.
Acero:
Yes. We’re going to. And, you’re going there, but I don’t want the competitor to go and beat me now. If you convince yourself you’re going to do it, at least I’m going to make it more difficult for Apple. In that case, putting all these impediments, [02:38:00] so that I’m not really going to help them compete with me more. I’m definitely not going to do that in principle for one year.
Hellrigel:
Wow.
Acero:
In the end, the company (Microsoft) settled, keeping me out of work for four months. I was unemployed. I had signed a piece of paper. [02:38:20] I couldn’t start working with Apple until late August of 2013.
Hellrigel:
How did they police that? Do they ask to look at your pay stubs?
Acero:
No, they didn’t. They don’t police it, but if we ever went to trial, you would need to get all the data, [02:38:40] so I just followed the book.
Hellrigel:
You were on vacation for four months?
Acero:
Yes, I was on a long vacation, and then for the following three months, I could not work on speech recognition.
Hellrigel:
Okay.
Acero:
My boss.
Hellrigel:
This means you were barred from working on the project Apple hired you to do for a total of seven months.
Acero:
Yes. But at least I was working for Apple because for that time [02:39:00] I didn’t know how long it was going to be. I didn’t know if it was going to be two weeks. It ended up being four months. The uncertainty was not fun because what if Apple changed their mind or if they wanted someone quick.
Hellrigel:
Yes, and what if they kill the product.
Acero:
They could kill the product. They might think this guy’s not coming, so let’s hire somebody else. It was kind of unsettling. But in the end, I joined. [02:39:20] The speech team had five or six people, so it was very, very small. They have hired and, in the meantime, the team was on the first floor. They told me, you are going to the fourth floor and don’t talk to these people. You can learn about other parts of Siri in the meantime. It probably would come in handy, which was good. We moved with the family. We bought a house and [02:39:00] wanted to remodel in the meantime. I was not working the eighty hours a week during those months.
Hellrigel:
You had to move from Redmond, Washington to California?
Acero:
We bought a house in Los Gatos, California, just south of Cupertino. We bought a house, started remodeling, found schools for the kids, and [02:40:00] did all these things. I wasn’t working as many hours as I used to, but that was in a sense good because I could spend some more time doing this.
Hellrigel:
At this point, how many kids?
Acero:
Two. Two boys.
Hellrigel:
Two boys. This move is a major shift for a whole family.
Acero:
Yes. [02:40:20] In fact, Donna had come down to visit to see what it looked like. We had at that point an eleven-year-old and a thirteen-year-old.
Hellrigel:
Wow.
Acero:
Donna’s sister came and spent [02:40:40] time with them, while we’re down there.
Hellrigel:
While you are starting out.
Acero:
Yes. And then when she knew this, told the kids obviously we’re just going out.
Hellrigel:
Go on a little vacation.
Acero:
We did some of these things. Microsoft would have a few trips, so her sister would come and stay with them. [02:41:00] It is not like this was the first time because we did that many times. When we came back, would they say dad we’re not moving. Maybe they could tell, the sixth sense.
Hellrigel:
Yes. They started to get some vibes.
Acero:
No. Just the older one who said, [02:41:20] “We’re not moving, are we? You have a job. How do you think of that? Do you?” So, he’s already suspecting something. He said, “Because if you do, I’m not moving.” Oh boy.
Hellrigel:
Your thirteen-year-old son was telling you. He’s wired.
Acero:
Pretty much [laughs]. [02:41:40] We’re not moving, unless we have a pool.
Hellrigel:
He wants a do not to move clause.
Acero:
I think that’s nice. I said, I see, you’re negotiating, maybe we can find that compromise.
Hellrigel:
So, a pool can entice him to leave.
Acero:
Yes, yes. In hindsight, [02:42:00] they were at an age. I hear from other friends, that when they’re in high school, it’s much more difficult, but one of them was in middle school and the other was still finishing elementary.
Hellrigel:
Right. They are young enough.
Acero:
They were at the age that was easier.
Hellrigel:
That would work for them.
Acero:
Yes.
Hellrigel:
Did they get a pool?
Acero:
[02:42:20] Yes, they did. That was part of the deal. You cannot break the deal.
Hellrigel:
And a new puppy dog?
Acero:
No, we do not have puppies because I’m allergic to everything.
Hellrigel:
I thought the younger son might want to have a piece of the “pie”, too, and ask for an accommodation.
Acero:
The younger son was more, go with the flow. I remember when [02:42:40] he first moved in, he said, “I miss the rain.” Are you kidding me. He said, “I miss the rain, Dad. It’s too sunny here.” So, I felt bad, but that didn’t last very long. Kids adjust.
Hellrigel:
I know you have to go to dinner.
Acero:
I have until well, like, probably twenty minutes. [02:43:00]
Hellrigel:
Okay. I know we may have to pick this up via Webex. If that’s okay.
Acero:
Yes, sure.
Hellrigel:
We can pick up with what you’ve done with Siri and all of that.
Acero:
I may actually have some time tomorrow before I go to the airport. I need to check the schedule. That’s possible?
IEEE Signal Processing Society
Hellrigel:
[02:43:20] Okay. I don’t know if it would be easier at this point to talk about IEEE and SPS [IEEE Signal Processing Society) and leave Siri for the next part.
Acero:
True
Hellrigel:
You’re probably joined IEEE [02:43:40], not AIEE or IRE.
Acero:
That’s right, IEEE.
Hellrigel:
It is the generational divide.
Acero:
Right.
Hellrigel:
If you were two decades older, you may have joined one of the predecessor societies.
Acero:
No. IEEE was founded, I thought it was in 1960.
Hellrigel:
The AIEE and IRE merged in 1963 to form IEEE.
Acero:
I joined in 1983. I was an undergrad. I was an undergrad in Spain [02:44:00]. I had no idea what IEEE was. In fact, I-E-E-E, it’s I triple E, I had no idea what that was, but they had a chapter—a local section, at the university. Somehow my professor, again said, “You should check [02:44:20] this out. This is like big deal.” I said, okay.
Hellrigel:
In Madrid?
Acero:
In Madrid.
Hellrigel:
So, you became a student member?
Acero:
I became a student member at that time.
Hellrigel:
Yes, I can check you in the system. Did they do any type of activities?
Acero:
They had some, but at that point, I just got one of the magazines [02:44:40] because this is when I was starting to do this project with speech recognition. So, it was at the same time. If you want to learn more about it, you need to subscribe to these magazines, these journals, the Transactions. This is where everything is happening.
Hellrigel:
Did your university have access to these publications? [02:45:00] In the U.S., many engineering colleges and universities have a library subscription to IEEE publications.
Acero:
I think that they had—
Hellrigel:
IEEE Xplore wasn’t around yet, so the libraries had hard copies.
Acero:
Right, right. They had access to the print copies. I became a student member; it was not that expensive. My professor there told me that that was a good thing, so I just trusted him. It wasn’t that [02:45:20] much money, so I said, sure. I’d do it. It is good for you; you learn more about the activities and so on. It is almost like a mentor tells you about that. Okay. That’s what I do.
Hellrigel:
You remained a member when you came to the U.S. for graduate school?
Acero:
I continued my membership all those years. [02:45:40] I’ve been a member all that time.
Hellrigel:
When did you start going to conferences?
Acero:
I was telling him that my first conference was in 1987 in Dallas. I was at Rice, and [02:46:00] ICASSP was usually this time of the year. Well, usually it is in April, so this (ICASSP 2023) is kind of late. It was probably in April, I think, March, April, May, but I thought it was April 1987. ICASSP was in Dallas, and it turns out that one of the professors at Rice that I had taken a speech processing class with [02:46:20] named Panos Papamichalis . He was SPS (IEEE Signal Processing Society) president at some point. He told me, “I am the organizing the General Chair of ICASSP 1987 in Dallas.” He was a part-time professor and [02:46:40] he was working at Texas Instruments
Hellrigel:
Okay. That makes sense.
Acero:
He said, “We are doing it in Dallas, sure. They had the papers, and they need student helpers. If you are a student helper, we’ll pay your registration [02:47:00] and hotel. Just come here.” This is to help out with whatever they tell you. It sounded good to me.
Hellrigel:
[Since you volunteered at a previous ICASSP and given your successful career], you could be an inspiration for everyone helping here at ICASSP 2023.
Acero:
Yes. Except we did more work then, because I was working the registration desk and those were the days of the five-volume Proceedings. [02:47:20]
Acero:
There were heavy things, and yes, one of my first customers is to pick up the Proceedings; Professor Oppenheim [laughs]. I had already studied that in my courses at Rice. I don’t think I wanted to wash my hands later on. [02:47:40] [laughs]
Hellrigel:
He (Alan Oppenheim) was a rockstar at that time, too. [He was so well-know because of his often used textbook.]
Acero:
Absolutely.
Hellrigel:
And you remember that meeting?
Acero:
Of course, I do.
Hellrigel:
You had his textbook?
Acero:
Yes. Yes. Of course, that’s how I knew him. He is like a rockstar coming here. I also handled the Proceedings. Professor Oppenheim is there anything else [02:48:00] you want, anything you need, do you need to find the room where you want to go? [laughs]
Hellrigel:
At that point, ICASSP was a smaller group?
Acero:
Yes.
Hellrigel:
In Dallas, you volunteered, you were a student helper, and you got to meet one of the big fish.
Acero:
[02:48:20] That’s right.
Hellrigel:
Then you’ve gone to most ICASSP?
Acero:
I went to 1988, to New York City. I was already at CMU and Rich [Stern] said, “Well, you don’t have any papers, but you come. You need to come and learn.” He paid for that. In 1989, it was in Glasgow. [02:48:40] I didn’t go, it was far away, didn’t have any papers. One thing is to go to New York City from Pittsburgh, and another thing is to go to Glasgow.
Hellrigel:
Yes. It’s a longer distance.
Acero:
So, I missed that. After that one, I have gone to all of them with the exceptions of pandemics. [02:49:00] Of course, I haven’t gone to the three recent ones that were virtual or hybrid. I am not counting them. And in 2003, ICASSP in Hong Kong was cancelled due to SARS.
Hellrigel:
Right. Yes, I forgot what was going on in 2003.
Acero:
So, I missed these four years. [02:49:20] [ICASSP 2020 to be held in Barcelona, Spain was virtual and ICASSP 2021 to be held in Toronto, Canada was virtual] Last year, [ICASSP 2022, held in Singapore,] was hybrid. Everything comes, so I missed ICASSP 1989, and ICASSP 1998, which I think was in Adelaide, Australia. [02:49:40] I missed 1998 because at that time my mom was diagnosed with breast cancer. I was planning to go, but obviously, first things first. So, other than that, I have gone to all of them [laughs]. That’s a lot. That’s a lot.
Hellrigel:
Is that a record? There should be a survey of who have gone to what.
Acero:
I am curious to see. [02:50:00] I don’t know if that would be a record. A lot of my colleagues here have gone to a lot of them, too.
Hellrigel:
Right. And I know India would be the fiftieth ICASSP. And K.V.S. Hari is already thinking of projects there. Maybe there will be an historical event for ICASSP similar to what was done for IEEE Signal Processing Society anniversaries. [02:50:20] The 2003 meeting was not held, but I do not know if papers were published. {The ICASSP people can figure it out.]
Acero:
[02:50:40] As Ray (K.J. Ray Liu) would say, this is my professional home.
Hellrigel:
Oh, yes. I was talking to Ray Liu, the 2022 IEEE President. Yes, he’s going to record an oral history with me. We have the [02:51:00] present collection [of IEEE Past Presidents oral histories].
Hellrigel:
Liu was pretty funny. He would tell us jokes at the Town Hall meetings (IEEE staff meetings), like he was doing last night. The IEEE president serves a one-year term, so the president changes so quickly, but each has their own agenda. Liu talked about making IEEE a member’s professional home, and I think this professional home idea is holding on [02:51:20]. Some of the other ideas don’t catch on, so they wane after the president’s term. In addition, Liu was the 2012-2013 President of the IEEE Signal Processing Society.
Acero:
Yes.
Hellrigel:
Do you go to [02:51:40] the other big conference?
Acero:
ICIP (International Conference on Image Processing) ?
Hellrigel:
Yes. ICIP.
Acero:
Well, ICIP is for image processing, so I never went there except as a volunteer. Once I joined all the committees and everything. The society (IEEE Signal Processing Society) has the Spring meetings that coincide with ICASSP [02:52:00] and the Fall meetings that coincide with ICIP, so I would go to ICIP, but not really to the sessions. I would just go there to the board of governors, the technical directions board, the membership board, or whatever was the board du jour. So, I went to the venues, [02:52:20] but not to the technical sessions.
Hellrigel:
This is your home away from home and one of your major publishing venues?
Acero:
Right.
Hellrigel:
In terms of IEEE, you’re an IEEE Fellow. I ask people, where were you when you [02:52:40] found out you were nominated and then when you became an IEEE Fellow?
Acero:
Well, my boss at the time, X. D. Huang, nominated me.
Hellrigel:
Okay. And you knew that?
Acero:
I knew that he was nominating me, and then he told me, yes, you got it. I got it the first time around, on the first try. [02:53:00] I had heard that first try happens sometimes, but a lot of the time it’s not the first try. It doesn’t happen on the first try, so you try one, a couple of times, or three, and then they give it to you. But I got it the first time.
Hellrigel:
I don’t know how they label it, but in my mind you’re the class of 2004. [02:53:20] I do not know if that is the official term. Now that you are on IEEE Fellow, what does it mean to you?
Acero:
Well, it’s the recognition from your peers that you’re the next step up. You are in the VIP section of the building, [Laughter] [02:53:40] of the IEEE building.
Hellrigel:
Yes, you’re with the VIPs. You are amongst the selected. Many IEEE Signal Processing Society members have become IEEE Fellows.
Acero:
Yes, which is surprising, but yes.
Hellrigel:
Right. [02:54:00] I don’t know if you can say it is a competition, but it is a competitive process. IEEE has thirty-nine technical societies and seven technical councils. I don’t know if these groups see it as a competition.
Acero:
When I was president [of the IEEE Signal Processing Society], I was in the board of governors, and one of the things that we had heard is that oh, we need to push because this [02:54:20] big, fat IEEE bureaucracy is not treating us right. We deserve more fellows than we’re getting, but that was not necessarily a statistic, it was an impression. So, I said, well, I’m going to look at the data. I like to crunch data left and right and up and down. I looked at it and realized, [02:54:40] actually, we are doing better than average when you normalize by the size. Now, of course, ComSoc (IEEE Communications Society) would have more. Well, guess what? They are a bigger society in comparison, but when you normalize by the size, we’re like near the top, so we cannot claim that we’re being discriminated. Now, it doesn’t mean we cannot try to get more, [02:55:00] but we should not go with the argument and say, hey, we are the bottom, and come on, this is not fair. But people didn’t know [the statistics], they had the impression that we deserve it more. Of course, the grass is greener, and you need to do this.
Hellrigel:
Well, yes, that is their perspective.
Acero:
So, I need someone to study and say, well actually, no. Now, what happened is that later I [02:55:20] found out there is a correlation with academics. Societies that are more academic, have more fellows. That is one of the correlations. Not the only one, but a strong one.
If you are in Power (IEEE Power and Energy Society), well, the percentage of members that are in industry, [02:55:40] or at least when I looked at it, I haven’t looked at it in at least eight years, is much, much higher. When I looked at it, it was 80 percent of the members of Power Society (The IEEE Power and Energy Society) were industry. Guess what? There are fellows from industry, but the percentage is significantly smaller. So as a result, when you normalize, hey, we have many more [02:56:00] fellows per capita than Power or other societies.
Hellrigel:
How about in terms of Ph.Ds., does the IEEE Signal Processing Society have more people with doctorates than the societies that are more oriented to industry?
Acero:
Well, yes, typically. Yes.
Hellrigel:
So that might correlate instead of accomplishment.
Acero:
Well, that’s true. Both academics [02:56:20] and Ph.Ds., those are more correlated. But a lot of the times when people go to industry and mine was an exception, I was in an industry research lab. So, in fact we published more than many universities. So, there’s not quite—yes, it’s industry, but not quite there in that sense.
Hellrigel:
[02:56:40] Yes. Yes, that’s why industry is—
Acero:
There are industry research labs and then industry. If you’re in industry, then you could argue what I’m doing now then it’s a much lower percentage. If I look at the percentage of people that I have in my team that are Fellows, it’s a very, very small number. [02:57:00] But when I was managing this, the group in Microsoft, probably like 30 percent or 20 percent.
Hellrigel:
Right. Versus if you worked for a power company or construction.
Acero:
Right, right.
Hellrigel:
The chances are you’d have a B.E. and maybe an MBA.
Acero:
Right, right, right.
Hellrigel:
So, that is part of it. [02:57:20] You did all that numbers crunching as the president of the IEEE Signal Processing Society.
Acero:
Yes.
Hellrigel:
Did anyone encourage you to get into that?
Acero:
Well, I started as a volunteer in the early 1990s, after I got my Ph.D., as a reviewer. The Speech Technical Committee needed reviewers. [02:57:40] That was the first time I went to Piscataway (the IEEE Operations Center is in Piscataway, New Jersey) because at that time the reviews were in person. At that point, they sent you hardcopies of papers in big envelopes. You read them and then you go there and discuss them and just vote for them in person at that point.
Hellrigel:
Quarterly meetings?
Acero:
No, once a year, for ICASSP. [02:58:00]
Hellrigel:
Once a year. Oh, oh, oh, ICASSP.
Acero:
Right.
Hellrigel:
Okay, you’re going to review paper proposals for the conference?
Acero:
Yes. Papers submitted to ICASSP. Today it’s all done online, but that wasn’t always true. That’s my first time in Piscataway.
Hellrigel:
Did they put you up at the Heldrich Hotel in nearby New Brunswick, New Jersey?
Acero:
I don’t even [02:58:20] remember which hotels.
Hellrigel:
But you stayed locally?
Acero:
I thought I stayed close. I don’t think it was in New Brunswick. I’ve been there many times later with all the board meetings. I thought somehow, we stayed closer. I don’t remember where.
Hellrigel:
But they’re not putting you up near the 3 Park Avenue office in New York City? [02:58:40] You’re not based out of New York City.
Acero:
No, no, no, no. We were in New Jersey.
Hellrigel:
You volunteer to review conference paper proposals.
Acero:
So, I was a member of the Speech Technical Committee. Kai-Fu nominated me. He had weight, so they accepted me [02:59:00] and I was a member. At some point, the chair—you worked with the chairs—was Raja Rajasekaran (P.K. Rajasekaran) who was at TI. He tells me we need to elect a new chair. [He said,] “Alex, I think you would make a great chair. I’ll nominate you, if you want to do it.” I said, okay, [02:59:20] so he nominated me and guess what, they elected me. I was the chair of the Speech Technical Committee. After I did that, I said, well, a lot of the times when you do that, you could go to the board of governors. Did I want to run for the board of [02:59:40] governors, and be on the ballot? I think I was pretty young and not very known, and of course there’s a lot of noise in these elections. I was not elected, but I was kind of, like, next. It turns out that one of the members [03:00:00] got voted to VP or something, one of the others left a slot, so they took the next and that was me.
Hellrigel:
Okay, so you were on the board of governors.
Acero:
I became the member at large on the board of governors. I was there and then eventually [03:00:20] I think this is when Al Hero was president I served in the board of governors when Al was president and Fred Mintzer from IBM. Then [03:00:40] they changed the bylaws and created this vice president for technical directions. [I was asked] why don’t you run for this? So, I said, okay. Now at that point, all of these positions were voted for by the board of governors, so they voted for me and I did that. I reviewed all the technical committees, [03:01:00] and then my term ended. So, we think we need to do something with industry owners and industry members. You work in industry. How about if we create an industrial directions liaison or something technical Committee. Hey, you’d be good. Why don’t you do that? I said, okay. [03:01:20] I’ll do that. After that point, to become president, you need to have served as a member of the board of governors. But in terms of that, if you look at this fine-print bylaws, you need to be elected and I wasn’t technically elected the first time, because I was just—did—
Hellrigel:
Slotted in.
Acero:
Slotted in because someone moved on. [03:01:40] Of course, it depends on how you read this because if I want—I got more votes than other people. votes and other people. So, I run. And so, you should be president, but you need to do that, you need to run for this. I ran, and that time I got elected through the front door. [03:02:00] Then after that I said, well, run for president. I ran for president and won, becoming president-elect and then president. So, I have been serving the Society for a long time, and of course after I finished my terms, I said hey, what is the next step, Division Director? Just like all of my predecessors have done. Right? [03:02:20]
Hellrigel:
This is Division IX?
Acero:
Division IX.
Hellrigel:
Yes. And that puts you on the IEEE Board of Directors?
Acero:
That’s right.
Hellrigel:
That must be a lot of fun.
Acero:
It’s kind of interesting. [Laughter]
Hellrigel:
Yes. Yes.
Acero:
As Society president I have attended the TAB (Technical Activities Board) meetings [03:02:40] which is like, this big room where everyone is disputing whether you put the comma here, or you put it there, or you change that into a period, or there’s heated debates on that and other stuff. But somehow, I just said, okay, I’ll run for this. [03:03:00] And there you go, I served on the IEEE Board of Directors, too. Now, after that, they said, well, the next step which is Vice President for Technical Activities. But at that point I said, “What, I’m not running.” I was fed up already. [03:03:20] While Microsoft research is very academic, and they favor these things, Apple wasn’t. It’s not like they said don’t do it, but they don’t encourage it. They don’t view it—and at that point I had had enough of the paperwork. So, I’m not going to do that now, and that’s a lot of work, too. I decided not to consider it. [03:03:40] Not to say I might have been elected, but I didn’t even run.
Hellrigel:
Right, right. Some IEEE members have told me, one of the biggest challenges is industry doesn’t have the time for its employees to do that, especially now.
Acero:
Right. You need to do it on your own time.
Hellrigel:
Yes. Yes.
Acero:
So that was the end. Then after a while, [03:04:00] of course you know of friends, and John Treichler, who I have known, was the president of the IEEE Foundation. He said, “Well, we need members.” He called me in, so that’s what I’m doing now. I’m on the board of directors of the IEEE Foundation.
Hellrigel:
Right. I know you have to run, but we can pick up [03:04:20] with the IEEE Foundation and Apple.
Acero:
Sounds good.
Hellrigel:
Sorry to take so much of your time.
Acero:
Yes. Oh no, it’s fun. This is a good exercise.
Hellrigel:
Yes, but let me take your microphone before you trip yourself up, sir. We are ending almost on time. We are [03:04:40] five minutes late.
Acero:
I need to see if tomorrow it may be possible, if it works for you, to do another session.
Hellrigel:
Yes, yes.
Acero:
My flight is at 10:30 p.m.
Hellrigel:
Okay. I’m around.
Acero:
I obviously will need to leave by, let’s say, 8:30 P.M. [03:05:00] at the most, or something around that time. I will have time after the session to talk.
Hellrigel:
Right, right. That works. I am recording Ronald W. Schafer’s oral history from 9:00 am to 12 noon, and then I’m around.
Acero:
Okay.
Hellrigel:
You could send me an e-mail and we can set the time.
Acero:
Yes. Sounds good. [03:05:20]
Hellrigel:
Okay. Enjoy supper, dinner. Yes, I think they all left for the day. I didn’t realize how late it is. There was a group that was going to meet downtown and I’m like, eh, we’ll see.
Acero:
It’s still not too late for dinner because that’s where we’re going. [Laughter]
Hellrigel:
Okay. Thank you.
Acero:
Thanks, bye. [03:05:40]
Hellrigel:
Bye-bye.
[End Tape 1, Begin Tape 2]
Interview, Part Two
Siri
Hellrigel:
[00:00:00] Today is June 9, 2023. I’m Mary Ann Hellrigel, Institutional Historian, Archivist, and Oral History Program Manager at the IEEE History Center [00:00:20]. I’m with Dr. Alex Acero, who’s an IEEE Fellow class of 2004. He is senior director at Apple with the Siri project. We’re at the ICASSP 2023 meeting in Rhodes, Greece and we are recording Part 2 of his oral history [00:00:40] When we left yesterday, you were just getting back to the world of Apple. You were brought in to work on the project called Siri?
Acero:
Yes [00:01:00], that was the name of the startup that was acquired by Apple and Apple decided to keep the name Siri. They like it, so they kept it.
Hellrigel:
You are brought in because you’re well-known in the field?
Acero:
Yes. They licensed the speech recognition from Nuance, and they figured [00:01:20] this is core for us, so we need to build this in house. We don’t have anyone because the startup was about twenty people, but it was mostly software engineers that did the application design. They just didn’t do anything on speech. They figured that they would bring me in to build the team.
Hellrigel:
We had talked [00:01:40] about your little bit of a bureaucratic annoyance when you were leaving your old company, Microsoft.
Acero:
Right.
Hellrigel:
You mentioned taking a four-month vacation. What did you do for four months?
Acero:
I had heard the phrase spring cleaning [00:02:00], but I had never practiced it for the twenty years I’ve been there [at Microsoft]. So, there was a lot of junk accumulated everywhere and this is my turn to do spring cleaning in the summer. I made lots of trips to the Goodwill store to donate stuff [00:02:20] and that’s what I did.
Hellrigel:
You’re spring cleaning your house because you are going to move from Washington State—
Acero:
To California. There was a lot of stuff that was accumulated. This is my time to do spring cleaning, so I did a lot of that. I made a few trips to the Bay Area to look [00:02:40] at houses and look at schools, while the kids were there in school. Then I started to improve my tennis game. I had been playing tennis recreationally when I was a kid, but I had stopped, and my wife had picked it up recently. She told me, why [00:03:00] don’t you do that? So those are the things that I did for the first three months. Now, at the beginning, there was nothing to do, and I was fine with that, but by the end, I was biting my fingernails. Donna said, “Don’t you have anything to do outside.” Plus, it was a little bit stressful because I didn’t know until [00:03:20] a few weeks before then when I could move.
Hellrigel:
Oh, yes, you had to find a new house and set up, and also find a school.
Acero:
Right, the school. We got the school, so we needed to move before the school starts, but I didn’t know when I would be allowed to start working.
Hellrigel:
Right. Legally you had to wait, but then it’s important to get somewhere by mid-August to start the [00:03:40] school year. Right, because at that point you have children.
Acero:
Two.
Hellrigel:
Two children, and they’re young. One’s going to start high school?
Acero:
Yes.
Hellrigel:
You have to show up on time for the beginning of the school year.
Acero:
Yes. We needed to get them space in the school, right?
Hellrigel:
Right.
Acero:
We’re going to be in temporary [00:04:00] housing. We’re looking for a house also, at the same time.
Hellrigel:
Okay, you haven’t bought the house yet, so temporary housing. You’ll rent a house?
Acero:
Right. The company, Apple.
Hellrigel:
The company will rent the house.
Acero:
Yes.
Hellrigel:
Are you going to send them to public school?
Acero:
Well, we didn’t even know what to do, so we found a private [00:04:20] school. The name was Harker. And by the way, that’s where Kate Schafer teaches.
Hellrigel:
Oh, okay. Kate Schafer, so this is Ronald W. Schafer’s daughter?
Acero:
That’s correct.
Hellrigel:
Harker school.
Acero:
Yes.
Hellrigel:
It’s in the Bay Area?
Acero:
It’s in the Bay Area. It’s a private school [00:04:40]. It has the reputation of being like the high-tech high school. There was no room. My wife had called, can we go there, say, no, no, what do you think. It’s already May, and admissions decisions we made months ago, and we are full. [00:05:00] I went there and somehow, I managed to get an offer for them to let our son. My wife says, well, you probably charmed the admissions officer or something, because when I went—when I talked on the phone, they said, absolutely none. So somehow, we got him in, and later on, the younger was, was in elementary, finishing elementary for [00:05:20] one year, but because he had a sibling, then he also went to Harker later.
Hellrigel:
Wow. Now, Harker probably has a lot of high-powered CEO type of people.
Acero:
Yes.
Hellrigel:
You are showing up as the big fish from Apple, and maybe that charmed [00:05:40] them, or they had someone drop out?
Acero:
It could be a coincidence. It’s really hard to tell. We’ll never know. We’ll never know.
Hellrigel:
Did they stay in the school?
Acero:
They stayed until they went to college.
Hellrigel:
Okay. So, is it co-ed?
Acero:
Yes.
Hellrigel:
Co-ed, and that’s quite [00:06:00] fortunate. They did not change schools again.
Acero:
Right.
Hellrigel:
Now, you’re moving from Redmond, Washington to the Bay Area. How difficult is it to find housing?
Acero:
It was not easy, but at that time it was 2013, so the economy was still barely recovering [00:06:20] from the 2008 big financial crisis. House prices were starting to go up. There was not much inventory, but we found one. In fact, it was the first house that we looked at. It was the first house that we looked at when Donna and I made our first trip. It [00:06:40] was not in the school district that we wanted, but if we ended up taking him to a private school, then it did not matter. Right? It was the very first house that we saw and we liked it, but the only thing we didn’t quite like was the school district.
Hellrigel:
You’re still in that house [00:07:00]?
Acero:
Yes.
Hellrigel:
Then your son got his swimming pool?
Acero:
He did. That was the bargaining chip that he used, so I had to get the pool.
Hellrigel:
So, they were very happy?
Acero:
Yes. But moving was hard. I think it was mostly hard for my wife
Hellrigel:
How come?
Acero:
[00:07:20] Because she said, well, you go to work, you are busy there, it’s the new thing, the kids go to school. Then I don’t know anybody here.
Hellrigel:
She has to start all over without an entry except maybe parents at school.
Acero:
Right. At the very beginning there was nothing. Over time, of course [00:07:40], but at the beginning I could see that that was hard. Fortunately, she had started playing tennis and she joined a tennis club nearby. And pretty quickly after that, she made friends, and she was playing tennis all the time. But the first few months [00:08:00] were a little bit tough.
Hellrigel:
Isolating.
Acero:
Yes.
Hellrigel:
Did her family visit?
Acero:
Yes, eventually, but not the first few months because we were actually in an apartment, and we were doing remodeling. We bought this house, but we made some remodels before we moved in. We bought it in August [00:08:21], just around the time that we moved, but we didn’t move in until March of 2014.
Hellrigel:
Oh.
Acero:
Six months. All these things take long.
Hellrigel:
Yes, that’s true.
Acero:
Once we moved in, it was the house, not an apartment. She had been settled. We had all been settled, so [00:08:40] at that point it was fine. But the first few months were a little bit hard.
Hellrigel:
What’s it like then moving into the new job? They finally let you on campus or in the office, and what’s the reception? You are a well-known person, but they also have a lot of well-known people there. I don’t know what it’s like for you to start a new job [00:09:01].
Acero:
Again, the first three months, basically my boss told me, well, you’re on a different floor, don’t talk to these guys, but talk to everyone else in Siri, learn, and then we’ll figure out after December is over, we’ll figure out what you’ll be doing. But roughly we know what you’ll be doing. He says, you can’t do it now. [00:09:21] So it was interesting to get back into the group, learn about other aspects of the—of the team
Hellrigel:
At Apple, you are going to be in charge of a research team?
Acero:
No. Apple doesn’t have research teams per se. At least, they didn’t. They don’t really. [00:09:40] It’s a combined R&D. The goal is to ship Siri, like to put the speech recognition, that runs in the data center at that time, and power Siri. That was the goal, to build the product. Now, of course, to do that, you need to do some research, too. But the goal is to ship a product. [00:10:00]
Hellrigel:
How many people were working on this product at that point?
Acero:
The startup was something like twenty people, but when I joined, my guess is there were probably like fifty people, something like this, total.
Hellrigel:
Wow. So, that’s a significant business priority [00:10:20].
Acero:
Yes. It was. The company was investing in it and more than doubled the team in a year, essentially, a year and change. It was decided, okay, we need to invest here if we want to do this.
Hellrigel:
This might be a stupid question, but Apple was known for computers, so, this is a major decision [00:10:40] to diversify the product line
Acero:
Well, the iPhone changed that. It was computers, but the era of the computers almost collapsed. The company almost collapsed in the mid-1990s.
Hellrigel:
Right. Yes.
Acero:
Then Steve Jobs came back in the late 1990s. Yes, Mac [00:11:00] kept going, but he started to do this iPod thing, and that’s what rescued the company. Then when the iPhone came, the whole thing changed. It used to be a company that made mostly computers to a company that makes mostly phones. So, it is a true consumer company. It was a big change. The iPhone launched in 2007. [00:11:20] By the time I joined, the iPhone was the number one product.
Hellrigel:
Yes, and it still is.
Acero:
Yes, it still is. So, at that point, it was just a consumer product. just add speech capabilities and language capabilities.
Hellrigel:
So, you’re [00:11:40] going to be the lead person?
Acero:
Right.
Hellrigel:
Lead researcher. How did you feel getting this new job?
Acero:
I was excited because again, as I mentioned, I’m interested in the science behind it, but I’m also interested in making a difference in the real [00:12:00] world. It’s just that for many years, the technology wasn’t quite there for most of what you like to do. At the time, wireless connectivity was improving, and the phones were getting more powerful. The speech recognition technology with these deep neural networks was making a big [00:12:20] disruption. All these planets and all the stars were aligning to the point they said, well, this is the time to do it. I was excited. It was a great opportunity.
Hellrigel:
A little bit before this, Verizon rolled out fiberoptics in my neighborhood and in some communities. [00:12:42] They prioritized where they were going to roll it out. Did fiberoptics improve the product? Was it better for you to work with fiberoptic lines or all wireless?
Acero:
Yes. It was mostly wireless for us. I’m sure [00:13:00] that those were using fiberoptics from backend server to backend server. But for us, I don’t think that was the key thing. It’s just that the very first iPhone, the cellular connectivity was not superfast. So, to do that, you needed to capture the audio and then transmit it. But there was a relatively [00:13:20] long delay. Then you need to process it on the server and get the results back, so it took a while.
Hellrigel:
Right. People may forget that you need the whole infrastructure so that Siri could tell you what to do.
Acero:
Right.
Hellrigel:
The bouncing back and that. Yes, a number [00:13:40] of people were talking to me the past few days about echoing and other challenges. You are developing this new product line, so did they give you a timeline? Did the bean counters give you a timeline and say, I want this on the market in X?
Acero:
No [00:14:00] because we already had a product. It would be more like replacing one thing that would be transparent from the user’s point of view. In fact, we didn’t make a big fuss about this at that time, but then eventually it was mentioned. We wanted to make it transparent. Most users just noticed, oh, now, it seems to be more accurate. That’s really all they will see. It [00:14:20] was under the hood, replacing one speech recognition engine by another.
Hellrigel:
How excited are you when it works?
Acero:
I remember the day that we wanted to switch over the traffic from the one server. The engineers brought me to this room and said, “Well, here is the big red button, [00:14:40] Alex, push it.” So, I pushed the button and all of a sudden, the traffic started going to our data center.
Hellrigel:
What’s the day?
Acero:
I think it was the summer of 2014; late summer 2014.
Hellrigel:
Did they take a photograph [00:15:00] of you [pushing the button]?
Acero:
Not sure. I think they brought some alcohol. [laughs]
Hellrigel:
They were going to let you do it.
Acero:
They did all of the work. I just pushed the button.
Hellrigel:
How successful was this?
Acero:
[00:15:20] It was successful in the sense that we were able to replace the engine in relatively little time and have it be more accurate than what we had at that time. It showed that building in house for a component that critical was important. Yes, we’re all excited that we did it in relatively [00:15:40] very little time, like, a year and change, right. The team had already started with this before I joined, but it was basically in a year.
Hellrigel:
That’s pretty quick. You continue improving it, so you’re still [00:16:00]—
Acero:
Right.
Hellrigel:
What other products are you developing?
Acero:
After we did that, my boss said, “Oh, that worked pretty good. Guess what, we still have the speech synthesis part, the part where Siri talks back, so maybe we can do that, too. So, I started another project on building a text [00:16:20] to speech system, as we call it, to essentially do the same thing. To do the voice of Siri. Along the way, there was the person, the voice talent, that Nuance had hired that was very popular and we talked about it. [00:16:40] She was replaced many, many years ago. When we built a new voice, we had to hire another voice talent, record her, and then build the voices. Actually, we recorded a female voice and a male voice. We did both. Then we built the technology [00:17:00] behind it and the voices. We also decided to build a studio. Normally, those things were subcontracted. Some big companies, Nuance and others, just hired somebody and said, “Well, you go and record there in your studio.” We decided [00:17:20] that it would probably be important if we have more control on the way that was done.
Hellrigel:
The old timers would say to control the acoustics, is this the issue?
Acero:
No, not just acoustics, but also like the direction.
Hellrigel:
Oh, okay.
Acero:
It says, no, no, don’t say that like that, say it like this. Kind of like what — [00:17:40] We hired someone from Hollywood. He was working in a studio. The terminology that he used is basically what they use in Hollywood, the producer, the director, the casting, all these terms that are applicable to making a movie. You are having someone like the performance.
Hellrigel:
Did [00:18:00] you hire a speech therapist?
Acero:
Not a speech therapist, but people that have—
Hellrigel:
An elocutionist?
Acero:
Yes. Essentially someone that says, you are too flat, come on. The weather is nice and sunny, just be more excited, right.
Hellrigel:
It reminds me of people making jokes about Eliza Doolittle in My Fair Lady. [00:18:20]
Acero:
Right, right.
Hellrigel:
The professor bet he could change her speech and pass her off as a lady and not a lower-class person.
Acero:
Yes, Professor Higgins. I love My Fair Lady. We watched it many times.
Hellrigel:
Is that what you are trying to do? You are trying to perfect a voice for the company’s product?
Acero:
Right. It’s not just reading for the sake of it, it’s reading and thinking. This is going to be used [00:18:40] as a material for Siri to speak back. We wanted Siri to have a particular personality, not just audio.
Hellrigel:
No monotone.
Acero:
Yes. I want someone that sounds helpful. How do you sound helpful?
Hellrigel:
Passionate, yes.
Acero:
There isn’t anything [00:19:00] that sounds helpful, but the directors can tell you. The director can tell you, “No, no, I’ve listened to many people, you’re not sounding helpful, you’re sounding irritated or you’re sounding —
Hellrigel:
Right. Not sounding warm and calming—
Acero:
Right.
Hellrigel:
Your pitch, your tone.
Acero:
Your voice is good, [00:19:20] otherwise we would not have selected you. But we still want you to be and to feel that way now.
Hellrigel:
That’s right.
Acero:
Yes.
Hellrigel:
Yes.
Acero:
So that’s why we control a lot more of the performance and the studio. We want to control the audio quality, and all of that stuff.
Hellrigel:
Are you doing this in English and other languages?
Acero:
Right. We started with [00:19:40] U.S. English because we were developing the technology at the same time. But obviously the goal is once we did that—and the same thing was true, by the way, with the speech recognizer, a year earlier. We did English, so well, now we need to do all of the other languages. It’s just, we couldn’t do everything at once.
Hellrigel:
Yes. It is it brands really [00:20:00] like the program, speech to text. I’ve never used it. I’m afraid that my New Jersey brogue will be misunderstood. I’ll have to test it. When I speak, I am concerned Siri will not understand my regional accent.
Acero:
Your English should be perfectly good for this dictation part [00:20:20]. By the way, when it means Siri, we also have a major different feature that is called dictation. If you open your iPhone and you open the keyboard, it is there. You may not have used this, but it is there.
Hellrigel:
I’m still learning the features of my iPhone, sir.
Acero:
If you click on this [00:20:40] at the bottom, here, there’s a microphone button.
Hellrigel:
Oh, yes.
Acero:
If you press that microphone button, you can just speak, and it will transcribe. So, for example, maybe I can do this.
Hellrigel:
Wow.
Acero:
I push the microphone button.
Hellrigel:
Yes.
Acero:
Then I speak and the system will transcribe what I say. [00:21:00]
Hellrigel:
Fantastic. And it’s simultaneous.
Acero:
Yes.
Hellrigel:
I have to experiment more. I’m a chicken.
Acero:
So, this is what we’re doing.
Hellrigel:
Yes.
Acero:
By the way, shortly after we had done that, I was traveling to a conference, and I was waiting near [00:21:20] the gate just for my flight to be called. This gentleman that comes and there was a cafe or something, so he goes there and orders some coffee or some food. The waitress leaves to process the order, and he brings out his phone and starts dictating, replying [00:21:40] to I swear, ten messages. Bam, bam, all by voice. It made my day. That was the technology we had been working so hard on, and there are some people that use it all the time.
Hellrigel:
Oh, my friends, even before the car phone, they would use [00:22:00] audio to text and maybe it was audio to email. They used to get their business work done when they were stuck driving on the Long Island Expressway.
Acero:
Yes, yes. Typing on a phone takes a long time, but here, you speak and boom, you’ve seen how fast it is.
Hellrigel:
Right.
Acero:
And it’s pretty accurate.
Hellrigel:
Oh.
Acero:
Like for some [00:22:20] people, maybe, a little bit less than others, but by and large, it is quite accurate. Even if you have to correct occasionally a word, it’s still much faster.
Hellrigel:
I liked my old slide phone for texting.
Acero:
You should try the dictation feature.
Hellrigel:
Yes, I’ll try that.
Acero:
So, we ship that also [00:22:40] at the same time that we ship that version of the new recognition in 2014. Of course, the technology has improved since then.
Hellrigel:
When you’re developing it, you’re testing the technology yourself? The developing crew is playing with it?
Acero:
Right. The way we do it is, we [00:23:00] collected some audio samples from our customers dictating. Then we transcribed them, and compared them with an offline data set, where we can say, well, now, if we run on a batch algorithm offline, not real time. And we can compare, we can see how many errors the system makes, and then we can finetune. [00:23:20] Once we are happy that the error rate is low on this offline data set, we put it in the internal population and Apple uses it, and then we can see how well it works and hill climb to improve it.
Hellrigel:
The IEEE Milestone Program is managed by the IEEE History Center. [00:23:40] There might be a number of technologies twenty-five years older, older that’ll eventually become milestones.
Acero:
Right. I know that they approved the Apple 1, Apple 2, and Macintosh, and they were working to have a dedication ceremony coming up. Of course, those are longer [00:24:00] than twenty-five. They are almost like the Macintosh is thirty-nine here. So now—
Hellrigel:
Right, right. Those are the rules. It has to be a technology twenty-five years old. The milestone is for the technology, not a person or a company.
Acero:
Yes, twenty-five. Right.
Hellrigel:
Milestones are appealing to Apple.
Acero:
Yes. [00:24:20] We don’t do it for the milestone.
Hellrigel:
Right, but the technology is eligible to be an IEEE Milestone twenty-five years afterward.
Acero:
We do it for users to just, one of the things that I learned when I first joined Apple, and the second time too they told me is, why is Apple so secretive? And the analogy that I was told is, “well, when you have kids and you [00:24:40] spend so much time preparing, looking for the gifts for Christmas, you put them under the tree, and then you see them, the little buggers coming from their bed jumping so excited. The adrenaline is pumping as they are just looking at what Santa left under the tree. This is what we do. We [00:25:00] put all these products out, and then we see people line up around the block to try to buy them.
Hellrigel:
Oh, I remember the lines then. I’m not an optimist or a pessimist; I’m a realist. Then part of me pauses and thinks, you’re doing what?
Acero:
The idea is, we do this, and customers are excited [00:25:20] just like the kids.
Hellrigel:
Oh, they are.
Acero:
In the meantime, we keep it secret. We don’t tell the kids that this is what Santa is bringing you. They may know if they wrote the letter, but that’s the kind of thing that surprises and delights the users.
Hellrigel:
I don’t know if you remember the [00:25:40] toy, the Furby?
Acero:
Yes.
Hellrigel:
One Christmas season, my father stood in line, and he went everywhere, even stopping along his route as a truck driver, checking if that toy was in stock. He had to find a Furby, and at another time a Cabbage Patch doll, for my youngest sister.
Acero:
They were in short supply at some point. I remember.
Hellrigel:
Right. [00:26:00] I don’t know if she still has her Furby. A flood in 1972 from hurricane Agnes destroyed some of the toys I and my siblings had, but my youngest sister was born in 1973, so a lot more of her toys survived. My dad and aunt both got her Cabbage Patch Dolls, and she still has them. As well as the Furby and other popular toys. It excited my dad to see his kids excited.
Acero:
I remember it was the hot toy of that year.
Hellrigel:
Yes.
Acero:
And they run out of stock because [00:26:20] the kids put them on their Christmas list. Well, we can’t find them.
Hellrigel:
Yes. Easter was the most important religious holiday, but Christmas, my dad really tried to make it special. My sister’s face lit up when she got the Furby. [00:26:40] It was much more advanced than my Woody Woodpecker that had the little recording in it and you pulled the string, said something. But the interactivity in and all. But Apple I can understand the desire to have the aura and the excitement.
Acero:
[00:27:00] Right. Milestones, I like them and everything, but the reason why everyone is there is to surprise and delight the customers. The wow, this is so easy to use. It feels like it’s designed all by Steve Jobs and Jony Ive the designer, which of course it isn’t. It’s just thought [00:27:20] through, every little thing.
When I joined the second time, they had an orientation class, and part of the orientation was showing me not a computer thing, not an iPhone, nothing. I’m shown two receipts. One is a receipt that you get in a place like Best Buy. [00:27:40] The other one is a receipt that you can get in an Apple store. The Best Buy receipt was this long. The back had, I don’t know how many ads, and in the front, the ink was kind of smudged and you could barely read it because it had been put in a back pocket. So, the question was, when was this purchased? That was the question. [00:28:00] It took a minute to try to figure it out. He said, well, now I’m going to do this. This is thick white paper, the printed receipt, just really sparse with just the information that you want. What was the product that was brought, the day, the price, and the store where you bought it. Something like that. And that was it. No ads [00:28:20] and no coupons. That’s the thing. Which one do you prefer as a customer? So, it was the experience, an obsession with the experience to be extreme, not just the product. The receipt, the box, and how you open it.
Hellrigel:
All the boxing and packaging. [00:28:40] I remember that.
Acero:
Everything, everything. So that permeates the culture.
Hellrigel:
And it’s still like that?
Acero:
Absolutely, yes. You talk to the designers, the marketing team, the designers, the packaging, and the distribution. They on-board the salespeople in the stores. It’s all the same. This [00:29:00] is more like, you want to tell them, you want to be helpful, but you don’t want to be in your face. So, you should look at someone and you can tell by looking at the person, does that person need help or would that person like help or does that person just want to be alone for now. And you can just tell. If you’re experienced and you know what you’re doing [00:29:20], you can tell. If someone is doing this, looking—then you don’t wait. But you can look at them, but sometimes—it’s like when you go to a bar, a restaurant, they just bring you the food. You’re eating, and then the waiter comes and says how is it going? What do you want me to do? What, I can’t hear [00:29:40]—my mouth is full. I just—
Hellrigel:
But you learn to read the customers, so their experience is different.
Acero:
Right. So that was applied to everywhere and it still is. That is the main thing that drives the company, not just the engineers and the researchers like me, but everyone. Of course, everyone does their own part [00:30:00].
Hellrigel:
You’re still working on Siri?
Acero:
Yes.
Hellrigel:
I don’t mean still. It’s a product line that just keeps going?
Acero:
Yes, we keep improving the accuracy of the recognition. We keep improving the quality [00:30:20] of the voice synthesis. Then there are other products, so we get spinoffs. We started putting it just on the iPhone, and then we added it into the iPad when the iPad shipped. Then there was a product called CarPlay, which is in many cars now—a lot of the new cars, at least, in the U.S. that you buy [00:30:40], have this screen in the dash where if you plug in your iPhone, all of a sudden it takes over the screen. Like normally, you buy a, you rent a car and you have no idea how even the radio works. Now, you plug it here and it’s the same thing as you are used to it. So, it’s very familiar. There was a point [00:31:00], years ago, not now, but years ago, shortly after a CarPlay launched that the car makers would tell us, we have the same car, the same color, the same engine, the same leather, non-leather with the only difference being one of them has CarPlay and the other one doesn’t. Well, the one that has CarPlay sells [00:31:20] much more quickly. The other one stays on the lot. And of course, it’s not just—like, if the car sells quickly, they make more money because they don’t have the inventory.
Hellrigel:
It’s not the same, yes.
Acero:
It doesn’t sit in there. I say, well, guess what? I’m going to bring more of those because those are the ones that I sell. Why? We were selling it because customers were asking for it. Why were they asking for it, because [00:31:40] you get there and it works, and it’s familiar. You understand the telematics in cars were not very user friendly. They work as well. So that took off. That’s a place where we put Siri as well. Then this little thing came out, the Apple Watch.
Hellrigel:
Oh, the Apple Watch.
Acero:
The Apple Watch, same thing [00:32:00]. We put dictation and Siri as well, in there. Then we have Apple TV. So, if you watch— And it’s on the TV also. Then the most recent device that we have is the Apple AirPod.
Hellrigel:
Oh, okay.
Acero:
All these little buds that you see, especially all the young kids, but many others too, wearing, [00:32:20] which is tiny. To pull, like any—it, it activates when you say the phrase, hey, Siri, but to make that happen, you—and that was like a tiny chip with a tiny battery. So, we had to work really hard to compress our algorithms in that little thing.
Hellrigel:
Now, this is a stupid question, but hey, Alexa [00:32:40], that’s the competitor’s project on Amazon.
Acero:
Right. What Amazon did is, they came up with an assistant that was a speaker, a smart speaker, and for that they were ahead. We came up with the HomePod a little bit later, but Siri was the first assistant in the market. Then [00:33:00] you had Alexa. Then you have others like Google Assistant and Cortana. Others came, but Siri was the first one.
In the case of Alexa, they came up with the first smart speaker, although—and, and one of the things that was particularly difficult is the far field speech recognition. It was recognizing [00:33:20] speech at a distance quite well. Although this is originally the topic of my thesis, doing speech recognition when the microphone is not next to you. In this case, it was even further away, right. When I was at Microsoft Research, we worked on a product called Kinect. That was very popular at that point for a few years [00:33:40], and then it kind of fizzled out. It was mostly a gaming device, but you could use speech recognition. It was kind of like living room distance, like what we call 10 feet distance or more. So, we shipped a product with Far-field speech recognition, but it was not on an assistant per se, it was just focused [00:34:00] on the games and maybe controlling the, the TV.
Hellrigel:
These products are amazing, and the secrecy is one thing. But there’s iPod, iPhone, iPad, so why not iWatch. Why does Apple [00:34:20] not call it iWatch?
Acero:
Well, that’s the one thing that I do the technology, I’m not—
Hellrigel:
Like the marketing - -
Acero:
I’m not in marketing—so I say— I don’t even bother. I say, well, there’s a marketing team, they know what they’re doing— They must have a reason.
Hellrigel:
Right. Because, yes, iPhone, so there’s probably a reason. I’d have to just look into it. [00:34:40] But, yes, then the TV and then the product line now producing TV shows, and all of this. What’s the future?
You’ve been there ten years now? How’s that going?
Acero:
It’s going well. On one hand it’s rewarding because we’re growing and we’re trying new technologies [00:35:00] at the same time. Oftentimes, it’s challenging with users because when people ask me, what is artificial intelligence, well, I gave the analogy of the difference between artificial intelligence and human intelligence is that the computer is intelligent at some things, and the human is intelligent at others [00:35:20]. And there are scenarios where they overlap, but there are areas where they don’t. For example, most people have gone to grade school, and they know how to do arithmetic. But if I give you a multiplication of two numbers of twenty digits each, and I just give you a pen and paper, well, how to do it, but there’s a good chance you’ll make one [00:35:40] mistake with that many digits, right?
Hellrigel:
Yes. Yes, and it will take longer.
Acero:
It’ll take longer and you may make a mistake. Whereas the machine does not make mistakes and does it in less than a millisecond. That’s a case where you say, oh, the computer, or definitely the machine intelligence, is much better than the human intelligence. There are other things that are more like perception [00:36:00] based, like communication through language where humans are generally better. There are some cases where you may challenge that. No, but by and large, they are better. So why can’t you do that? I can do that. My kid can do that. So that’s a case where we’ve been working on this for decades, and it’s been difficult because the expectations [00:36:20] of the users are you’re going to do just like my other friend, my son. And it’s very difficult because you almost need to replicate human intelligence. Not quite necessary to do the transcription or the—or part of the understanding, but a lot. You need to replicate a lot of it, right. And that’s really [00:36:40] challenging.
Hellrigel:
Right. And there’s some things that people might expect it to do. Like, oh, we can just have the computer do that. Like say, you can—like say, psychological counselors, they have to know how to read the person, their physicality and all that. And the computer, you could say to the computer, [00:37:00] I feel X, Y, or Z, but it can’t read your emotion.
Acero:
Right, right.
Hellrigel:
But there might be people in the general public saying, oh, we could get a computer to do that. It reminds me of using assembly lines of robots. There was this high-end china manufacturing company [00:37:20] and they thought that they could just get robots to come and make the plates and move them and all that. But what they found was that the human touch was softer and it was going to take a lot more to perfect the robots. They wound up getting dumpsters and dumpsters of broken china. [00:37:40]
Acero:
[Laughs].
Hellrigel:
Maybe the, the assembly line can move a box, but it doesn’t have the fine dexterity for totally automating china manufacturing.
Acero:
And of course, the technology will improve, but it is a difficult problem.
Hellrigel:
Right. And it’s not cheap.
Acero:
Yes. And it’s a matter of like, humans thing, well, I do that, why cannot your computer [00:38:00] do that, well, not that easy. Replicating or trying to mimic, in some sense, human intelligence is very difficult.
Hellrigel:
It is. I don’t know if you want to say any more about Apple.
Acero:
The future of course is hard to tell, and Apple, of course, is very secretive.
Hellrigel:
Right [00:38:20].
Acero:
I can tell you the things that we’ve shipped. I can tell you things that we’ve published, but not what we’re working on. But—oh, by the way, the latest product was announced on Monday at the Worldwide Developers Conference. It is the Apple Vision Pro, which is these ski goggles [00:38:40] lookalike.
Hellrigel:
Yes the headset that looks like ski goggles for the experience.
Acero:
Augmented reality. But it has Siri. So, we put Siri in that as well.
Hellrigel:
Okay. Yes. One time at the IEEE office they had that virtual reality thing, and they had two sets in, I forget, it was 2016 [00:39:00] or 2017. It was around the time I joined the IEEE staff. It was a weird experience for me.
Acero:
That’s the kind of thing that is really hard to tell whether it will take off or not.
Hellrigel:
Right. I know the virtual reality. My nephew is into his gaming system and all that.
Acero:
Yes, yes.
Hellrigel:
My nephews and niece are having fun with their game systems and virtual reality games. [00:39:20] It’s the latest product. So, it’s a happy day at Apple then?
Acero:
Yes, yes. Well, that team has been working hard for a number of years. So [laughs], and this is just the first the announcement is not shipping for a few months. And it’s the first generation. If you look at the first iPhone and what it looks like now, there [00:39:40] is a big difference. The first iPhone, for example, only had messages, phone, Internet and the browser. That was it. Now, all the app store, that came later.
Hellrigel:
Oh, yes.
Acero:
That was not in the first version. Then all the apps. It didn’t have a camera, like 3G, 4G, 5G, make it faster, more powerful. These things happen. The first version was already very valuable, but it was just showing the beginning, the tip of the iceberg there as the developer community adopts [00:40:20] it and exploits it, and we learn what works and what doesn’t. There’s more technology built in this is a new category, right?
Hellrigel:
What are we up to now iPhone 16?
Acero:
Well, iPhone 14, but iOS 17 was announced, but yes, it’s a little bit confusing. [00:40:40]
Hellrigel:
Okay. The iPhone 14 is what people can buy. Then is their planned obsolescence? Does Apple plan to put out a new phone and have people buy a new phone like every year, every eighteen months?
Acero:
That’s really up to the user. What Apple does is maintain [00:41:00] and give updates to the operating system for, I think it’s something around seven years. A phone that was bought at that time will get an updated version of the operating system. After that, you won’t get the update, but that doesn’t mean it stops working. the upgrade cycle [00:41:20] used to be faster when the technology was going more quickly. Now, maybe some people change it every year. Some people change maybe three years. But you can see some of the older iPhones around, too. They work and sometimes you may need to replace the battery [00:41:40] because the battery doesn’t hold the charge after so many years. Right?
Hellrigel:
Right. What happens to your old iPhones? Do they go to the less developed parts of the world, or?
Acero:
They used to do that. I don’t know what happens to the original 2007 iPhone. I’m not really sure. [00:42:00] But yes, at first, they get sold in other parts at a discount.
Hellrigel:
And refurbished, yes.
Acero:
And refurbished. But a lot of times, it’s hand-me-downs. The father and the mother buy them and say, well, I’m going to buy the new one. Little Joey is asking for it. Oh, here, hey, this is for you, little Joey.
Hellrigel:
Right. Yes. That’s what happens [00:42:20] in many families.
Acero:
Yes, of course. So, it’s not just for developed countries, but my kids. I’m not going to give you a new one, but if I’m going to buy a new one and I can trade this in, so I know whether I’ll give it to you. [Laughs]
Hellrigel:
Right. Right. It is like an automobile.
Acero:
Yes, basically.
Hellrigel:
The money you get [00:42:40] for a trade in might make the phone more valuable as a hand-me-down.
Acero:
If you have a kid that can’t wait, you either buy a used car or you give him your hand-me-down, and they are happy. That’s what my kids did. When we moved down to California, I had a Volvo convertible, and now, [00:43:00] it has its years. It’s a 2007. But the kids, when they come and visit, love to drive it. Take the top down, it’s so cool. All red. [laugh] It doesn’t have all the high-tech gadgets of the new ones. You probably don’t want to take it on a very long trip, but just to drive around when they are visiting us. [00:43:20]
Hellrigel:
They learned to drive in that car?
Acero:
Yes, of course.
Hellrigel:
They graduated from the Harker School?
Acero:
Yes, Harker.
Hellrigel:
And went off to college?
Acero:
Yes.
Hellrigel:
Any engineers?
Acero:
The older one followed my footsteps [00:43:40] and ended up also at Carnegie Mellon, in engineering, but decided not to do electrical engineering or computer science. He decided to be a material science engineer. He likes to touch things. He got his master’s degree in December. He’s been working already for six months, [00:44:00] building lasers. He goes and puts all the phosphorus in the furnace and adds all of these rare earth materials, I’m not really sure what else he does, and out comes this laser [laugh].
Hellrigel:
And his name?
Acero:
Nicolas.
Hellrigel:
Nicolas.
Acero:
Without an H. [00:44:20]. He’s working for a company called L3Harris. It’s the merger of L3 Communications and Harris. The facility is in Hilton Head Island, South Carolina. [00:44:40] We visited and there were something like thirty-six golf courses on the island, which is not that big. And it’s a resort town. It’s like hotel after hotel, resort, resort, one after the other, at the beach, and then a laser company [laughs]. Is like—
Hellrigel:
That’s strange.
Acero:
It is very strange.
Hellrigel:
Well, yes. [00:45:00] Maybe someone retired and started that up the laser company?
Acero:
No. So, what happened is that that was L3Harris bought this company and that company was built by—it was a startup, that someone started in Ohio. And somehow, at some point, they say, what, I’m going to—I don’t like Ohio, I’m going to move somewhere else, it’s just my startup so I’ll just move down there, I like it here. Then L3Harris bought it, and they kept it there. L3Harris is a big company. It’s one of the big defense contractors after Boeing, Raytheon, Lockheed Martin and—I forgot one more, but it’s like up there, Number 5—Number 6.
Hellrigel:
Does he play golf?
Acero:
He doesn’t, but I say, this is your chance to play golf because maybe you’ll find another job somewhere else. Here, there is a golf course, so yes, I’ll pick it up. I’ll just do it when it’s not high season and It’s expensive.
Hellrigel:
At Stevens Institute of Technology where I taught, I had to do many things and one of them was teach PE, physical education. I actually taught rugby, but golf was the course that always had a waiting list. It made sense because in the old days so much business was done on a golf course. They learn to play golf, at least, not whack it too brutally.
Acero:
Yes.
Hellrigel:
And your youngest is—
Acero:
My youngest son thought that all of these engineering things, are too applied. He’s a true scientist. So, he’s studying physics. He just finished his sophomore year at Carleton College in Minneapolis. A little close to Minneapolis. It’s really cold. Now, he’s spending the summer at Cornell University. He’s doing [00:47:00] a research experience for undergraduates in a program that the NSF funds to learn how to build particle accelerators.
Hellrigel:
Okay. Accelerators. Superconductivity.
Acero:
So, well.
Hellrigel:
Superconductivity.
Acero:
Yes, yes.
Hellrigel:
I have recorded oral histories for the IEEE Council on Superconductivity.
Acero:
This is his first week. I haven’t talked to him. He flew there last Sunday. [00:47:20]
Hellrigel:
And his name?
Acero:
Marcos.
Hellrigel:
Neither of them became a portrait painter or a sculptor yet?
Acero:
No, no.
Hellrigel:
You and your wife must be happy that they’re on their [00:47:40] way?
Acero:
That’s right. They’re doing great. I visited Nicolas a few weeks ago because the IEEE awards ceremony took place in Atlanta this year. So, I went there as part of the IEEE Foundation and also because Apple [00:48:00] is the sponsor of the Kilby Medal (the Jack S. Kilby Signal Processing Medal).
Hellrigel:
That’s right.
Acero:
It turns out that this year the recipient of the Kilby Medal was one of my former Ph.D. committee members, José Moura.
Hellrigel:
José Moura.
Acero:
He was there.
Hellrigel:
I saw his plaque down there.
Acero:
Yes, yes. I was there and then [00:48:20] my son drove from Hilton Head to Atlanta. He came for that symposium, all the talks, and the ceremony. Then we drove back together. I worked from his apartment for a week before we all flew to Pittsburgh for the graduation and commencement ceremony.
Hellrigel:
That is great. At least, he got to walk. [00:48:40]
Acero:
That’s right. He finished the course work in December, but the ceremony is not until May. So, we did that. I spent a week with him and I was worried how is that going to be, but we got along great. I was happy to see that he was doing so well.
Hellrigel:
This was a bachelor pad with you and your son?
Acero:
Yes, yes, it was good.
IEEE Foundation
Hellrigel:
And [00:49:00], that’s fun. You mentioned the IEEE Foundation. How did you get involved with the Foundation?
Acero:
Actually, I started when I was in Microsoft research. Along with two other colleagues, the person that hired me, Xuedong Huang, [00:49:20] and another researcher there, Hsiao-Wuen Hon, we decided to write a book about speech processing. Somehow, it eventually became a 1000-page book. It took a long time, one year, [00:49:40] to finish the first thousand pages and then longer to polish it and clean it up. We published it in 2001 and we decided, well, we’re not going to make any money out of this. We did it just because we wanted to do it. There were some speech books, but they were old, [00:50:00] so we did something up to date. We said, what are we going to do with the money? If we don’t want it, we’ll just donate it for student travel, so we started the Spoken Language Processing Grant to pay the travel costs for students that had a paper at this conference. [00:50:20] The mechanism to do it they told us, oh, you should do it through the IEEE Foundation. [We said the] IEEE what? Foundation, what is that? So, that’s how I learned about the IEEE Foundation back in 2001. I didn’t know [about the IEEE Foundation] until then. I would talk to some [00:50:40] of the folks there, the staff, and, and the president. John Treichler was president [of the IEEE Foundation].
Hellrigel:
Yes. Yes, John Treichler.
Acero:
I knew Karen Galuchie from those times, but then at some point, John tells me, hey, we’re looking for new members. I had finished my term on the IEEE Board of Directors, and [John] [00:51:00] said, well, you want to come here. I already knew at least what the IEEE Foundation was because of our book. We just donated the royalties.
Hellrigel:
Right. Recently, I interviewed John McDonald and recorded his oral history.
Acero:
Yes, and he’s also an IEEE Foundation board member.
Hellrigel:
Yes. Previously, I recorded oral histories with a few others associated with the IEEE Foundation. John Treichler is on the list of people to record an oral history. A few years ago, I recorded Leah Jameson’s oral history for the IEEE Foundation’s history project. [I also recorded an oral history with Richard Gowen, the 1984 IEEE President, focusing on his work with the Foundation.]
Why bother to be a member of the IEEE Foundation’s board?
Acero:
Well, I knew what the IEEE Foundation was. [00:51:40] John asked me, and I said maybe I can help a little bit and I will learn something along the way. At least I knew some of the activities that the Foundation does because of grants, and I knew that [IEEE] Awards was involved, too. Then I got to learn more. It’s, I think, Ray Liu, who you probably also talk to [00:52:00] says IEEE is a professional home.
Hellrigel:
Yes. Yes. K. J. Ray Liu the 2022 IEEE President’s motto is “Make IEEE your professional home.”
Acero:
It’s been like that for me. I don’t know for everyone, but definitely it was for me since I was a student in Madrid. I’ve been a member for now, it’s probably almost forty years. And I owe.
Hellrigel:
Soon you will have “life” status in IEEE.
Acero:
When I turn sixty-five. I already have more years, but there’s 100 years in total which I already do. But sixty-five, so four more years. I have been a member for so long and it helped me along the way, so if there is an opportunity to help now [I volunteer]. I can’t say yes, every time that someone tells me because [00:52:40] that’s a volunteer role and I don’t have [unlimited time but] I can spend some time.
Hellrigel:
Right. You do not have endless hours to volunteer.
Acero:
If I retire, maybe I can do a lot more, but now I have to say I’ll take one of these committees, one of these boards. I cannot take two or three, but I can take one.
Hellrigel:
Right. Since you have been spending more time as a volunteer with the IEEE Foundation, are you doing a little less with the IEEE Signal Processing Society? [00:53:00]
Acero:
Yes. Yes, in general, my volunteer time with IEEE is almost limited so I used to do a lot with Signal Processing [the IEEE Signal Processing Society]. I was director, board of governors’ member, vice president technical directions, director of Industrial [00:53:20] relations, president, and past president, you name it. There’s a lot of years so once you add them up. That was very inward focusing on the society. Then when you are a president already, you go to Technical Activities, and you start seeing other parts of IEEE that I didn’t really [00:53:40] know and then going to the IEEE Board of Directors as Division Director IX [Signals and Applications]. So, from that point on, it was—I say—of course, I still keep track of what’s going on in Signal Processing (IEEE Signal Processing Society) because it is my home Society. But now, I don’t spend as much time. I come to ICASSP all the time, but [00:54:00] other than that, I don’t do the committee work. There are other people in the Society.
Hellrigel:
During the conference, I have been talking to people excited about marking ICASSP’s fiftieth anniversary. I said maybe the organizers can figure out who’s still attending and has anyone attended all of them. It could be a nice trivia question, who has attended the most ICASSPs.
Acero:
I probably must have attended over thirty by now because my first one was 1988 and this is 2023. I missed the three for the [00:54:40] pandemic recently, and the one in 2003 in Hong Kong [was cancelled]. then I missed, I think the one in Adelaide, Australia because that’s when [my mother was ill]. I’ve gone to all of the others, so I’ll try to not break that tradition.
Hellrigel:
So, you [00:55:00] didn’t do the virtual and the hybrid ICASSP meetings?
Acero:
No.
Hellrigel:
Do you listen to webinars or—
Acero:
I don’t have time. I find myself that there’s so many things. I need to decide what I do because I have lots of choices and I know I cannot do all of them, so I have to choose.
Hellrigel:
Right. Do you listen to podcasts while you’re driving? [00:55:20]
Acero:
I listen to the news. I listen to NPR news, Apple News Today, or World News Tonight. Those are the ones that I tend to listen to. Occasionally, if I’m done with the news, and it’s a little bit longer drive, then I just play some music.
Hellrigel:
You’ll listen to the news and jam the current events into your [00:55:30] brain while you are driving.
Acero:
Yes, there’s so much content out there and I don’t have time so I have to be selective and pick a few things that I really want to do and say, well, this I like to do, but I just can’t.
Awards, reflections, closing remarks
Hellrigel:
Now, in terms [00:56:00] of IEEE you’ve received a number of awards and you are an IEEE Fellow, which is important. In 2013, you received the IEEE Signal Processing Society award for the Best Paper in Machine Learning for Signal Processing.
Acero:
Yes, the paper award for 2012 or 2013. I received the Society Award, the highest honor.
Hellrigel:
Yes, you were recipient of the SPS Society Award, but I found conflicting dates claiming both 2017 and 2018. Who told you that you received that award and where were you when you were notified?
Acero:
I was nominated. Normally, when you get this, I think most of the time you know because somebody calls [00:56:40] to say I’d like to nominate you for this.
Hellrigel:
They asked you first?
Acero:
Typically. Let’s say if it’s like a paper award then you have no idea. The evidence, the paper, is there, so they don’t need to ask you something about yourself because when it’s a personal award they need to write something to say, well, this is what this person has done and [00:57:00] maybe someone that nominates you knows about you.
Hellrigel:
They’ll need to write the nomination.
Acero:
But they don’t know everything, and they want to write a strong nomination, so they ask you to at least revise the draft, polish it, give them bullet points or a resume or something.
Hellrigel:
Who nominated you for the award?
Acero:
Let [00:57:20] me see. I’m trying to remember because there were more than one.
Hellrigel:
That must have been a great honor.
Acero:
Well, that’s of course humbling and exciting that someone thinks to [00:57:40] nominate us especially in industry because what you tend to see is that the majority of these awards tend to go to academics. Perhaps because they publish. They tend to publish more. They go to more of these meetings. It’s more common in academia to get these awards.
Hellrigel:
How do you measure [00:58:00] accomplishment? Do you measure by the number of papers, the number of books, the number of patents, the number of successful products marketed, etc. In industry publications might be less important.
Acero:
Right. If you work in industry, you’re doing. I published papers when I was in research, but not as many because once you do management, you don’t [have time]. I was telling the folks there the last few years that I was managing research, don’t put my name there [00:58:20] because I really haven’t contributed anything. Of course, in the last decade I haven’t written any papers, but there is a lot of work and important stuff that is not as visible as a publication. For someone that is judging, that’s the problem of people in industry, right? [00:58:40]
Hellrigel:
Yes. Oh, how do you like management?
Acero:
Well, as I told you the other day, I didn’t start with the right foot. But then I enjoyed it because it’s a different skillset. You need to try. It’s a people business. Software is a people business, so contact sport. [00:59:00] You need to understand what drives that person because the best thing you can do is the sweet spot. If I give you a problem that you’re excited to work on. If you’re excited to work on it, you’re going to do a better job than if you’re not excited to work on it.
Hellrigel:
Exactly.
Acero:
If you’re excited, and you do a good job, that’s good [00:59:20] for me, so I want to find the problems that are there at the intersection of problems that we like to solve and that you are excited about. What is that intersection? Maybe I have a great problem, but you’re not excited to work on it, but she is. Okay, I’ll give it to her then. I’ll give you something else.
Hellrigel:
Right. So fit the person’s skillset, interests, and temperament to a project.
Acero:
Right [00:59:40].
Hellrigel:
Does Apple give managers training?
Acero:
Yes, of course, and I had taken training when I was at Microsoft research, too. Then it’s always good to have these refreshers. As I said, I have the left brain, the part with the math, and there is the right brain [01:00:00], which is more the social aspects. So, it’s always good to have courses to refresh, exercise the right brain, and not just the left one because the left brain is exercised all time. Like we need to go to the gym, you need to exercise, you need to do that [for the brain], too. So, we do that occasionally.
Hellrigel:
[01:00:20] Apple is such a well-known company and you have become well-known, so do people see you and try to give you their resume?
Acero:
Yes, of course, but I’m sure I’m not the only one, but yes. You walk down the halls and lots of people stop me, hey, Dr. Acero or something, and I smile thinking [01:00:40], they basically kind of hinted that we’ve met before, but I try to remember people - - I don’t know, and just smile. I don’t know who this person is.
Hellrigel:
It’s funny I ran into somebody at this conference, and I said I want to introduce myself because we’ve met. They seemed puzzled and said where. I said, yes, we met, [1:01:00] I am the person who recorded your oral history.
Acero:
Oops.
Hellrigel:
They are very well-known, so maybe they are overwhelmed with people approaching them.
Acero:
Yes.
Hellrigel:
In addition, during the past three years with COVID, you met a lot of people virtually, but it’s different when you meet them face to face.
Acero:
Right, right.
Hellrigel:
[01:01:20] So, it was kind of a joke, and they are a great person. When I do these oral histories, I have a backlog list of people recommended by others. It is just me. Many of the potential oral candidates are “big fish,” so it’ll get there. [01:01:40] It’s just amazing.
Another question, you have 160 patents?
Acero:
Yes.
Hellrigel:
What do patents mean to you?
Acero:
Most of them are patents I filed while I was at Microsoft Research. [01:02:00] Oftentimes, we wrote a paper, submitted it to a conference, and thought well, there may be something here that we can patent. When we submitted the manuscript to the conference, we also sent it to our internal lawyers. Normally, [01:02:20] they looked at it and said, okay, that’s a good way to understand technology. Then they had a meeting where they asked a few questions and then they did like a translation from our paper and our conversation. Then they wrote it in a way that is essentially the same, but with the legalese language (laughs).
Hellrigel:
Did you have to run your papers past them before you published them? [01:02:40]
Acero:
That’s something that made Microsoft Research unique. The other day I mentioned the reasonable person principle. We trust you and you just didn’t need to get your manager’s permission, but managers would [01:03:00] give permission very liberally. So occasionally, he may say, yes, let’s get a second opinion. But by and large, you just do it, and you publish it.
Hellrigel:
This might be a delicate question, but does Apple encourage publishing? [01:03:20]
Acero:
It doesn’t encourage it in general.
Hellrigel:
If the culture’s a little bit more secretive, publishing might not be emphasized or encouraged.
Acero:
Right.
Hellrigel:
They don’t like to let the secrets out.
Acero:
Well, and I think that’s part of it, but I think another part of and why when I—why I joined, I—one of the things that I wanted to do is, hey, I don’t need [01:03:40] to publish anymore, I’ve published enough. But the people that I want to hire still want to do it so I wanted to persuade the executive team and the senior vice president to do that and I got push back. And one of the things, it’s reasonable to say, well, I want the employers to come here thinking I’m going to make the product [01:04:00] better for my users. I don’t want them to come here and say, I want to see what I can do to publish—to get this paper accepted.
Hellrigel:
Folks think about their reputation and building their reputation. Yes, that is important.
Acero:
And now, the negotiation that - - well, if you make the product better and there is something new that you think is worth publishing then [01:04:20] I will let you do that. So that’s kind of the implicit deal.
It’s not black or white, but it’s kind of like along those lines, meaning you come here to improve the product, if along the way—because the writing of the paper itself doesn’t take that much time. What takes the longest is doing the, the research. So [01:04:40] it is extra time, maybe a week or two, but compared to maybe a year, this is not a big deal. So, say, okay, fine, you can if you want to do that, that’s okay. It’s not necessarily encouraged, but at least it was not discouraged. I had to work hard at the beginning to convince them.
Hellrigel:
The cultural difference.
Acero:
Yes. [01:05:00] My boss told me when I tried to do that at the beginning, he said, well, you need to get some internal credibility first before you bring it up. I said, okay, wait a year or two until you ship some of these things. Okay, I’ll do it. They say well, I can trust that this person is not coming here just to publish and not help us. I already shipped a few things that [01:05:20] made our products better, so, okay, I will trust you.
Hellrigel:
You’ve patents, 250 technical papers, a textbook, this book in language processing, so are you writing anymore books? What’s going on?
Acero:
Now, we often get requests because this is like a [01:05:40] twenty-two-year-old book at this point and it was used as textbook in many graduate courses all over the world. Now, twenty-two years in our field is a long time. It’s been out of date. There are some chapters that’s still valid, but a lot of the content [01:06:00] is outdated so often, they will say when are you going to write another one?
Hellrigel:
Second edition.
Acero:
And the answer—second edition, I say well, we don’t have time. I was just talking to Xuedong Huang, who I saw on Monday. I came here on Sunday, and he was here on Monday, and I ran into him. We’re staying in the Sheraton, the same hotel. We were [01:06:20] having breakfast, and I said I was planning to go to Old Town, so why don’t we go and just walk around. He said, well, that’s where I was planning to go, so let’s go together. While we were doing this, we said, hey, maybe when we retire, we’ll write the second edition because now, we just don’t have time.
That is all I remember when we wrote that book. [01:06:40] My first son was born and when you have a little one, you’re not sleeping very much, so then I took advantage of this to start working at late night. My wife would stay, when I was working during the day [01:07:00], she helped to stay up and then when I came home, I say, well, now, you take a nap when you can. Well, when the baby wants to eat, we all need to get up, but the rest of the time, you go and sleep, I’ll be with the baby. So, I put him in the bouncing chair, slept a little bit, and watched him while I was working on the book. It took many, many hours, I don’t even know how many, and [01:07:20] a lot of weekends and evenings. That’s a big commitment. I think we would like to do it, but we need to maybe wait until we retire and have time.
Hellrigel:
Maybe you have to bring in the co-author and other author.
Acero:
That’s another possibility. Bring another author and say, well, you can have the name and then you do all the delta work (laughs). Which could be substantial [01:07:40].
Hellrigel:
Ron Schafer said that he wrote the book with Alan Oppenheim often after the kids went to sleep at the kitchen table
Acero:
Yes, those are the things that you need to do. I read and I study, Oppenheim and Schafer and then Rabiner [01:08:00] and Schafer.
Hellrigel:
Yes, Schafer wrote twelve books and the last one, in 2010, was the third edition of the second book with Alan Oppenheim. The students that were there yesterday at the meet and greet. You would have thought they (Alan and Ron) were rock stars. [01:08:20]
Acero:
Absolutely, they are.
Hellrigel:
The adulation caught Dr. Ron Schafer off guard and the students were just gleaming and so excited. I said to Ron, well, it’s probably not often that they get to see the people that wrote their textbooks.
Acero:
I remember a quote [01:08:40] that Al told me. The magician makes complicated things simple, like I do this, and it disappears. Well, to make it disappear, I had to do a few more tricks. And a good scientist needs to make the complicated simple and that’s why they are masters at simplifying. We try to do the same thing [01:09:00] because it’s easy to read a paper and the science behind it is often impressive, it was published, but it’s very difficult to understand. And then if you can say well, let me—this is not necessarily simple, but I’m going to simplify it so that you—at least is understandable by more people. And that has its own magic [01:09:20]. It’s not that easy. You need to like, truly abstract it to the point, say well, how can I explain it to more people even if they don’t understand all the details so at least they understand the concepts. Now, they can go and try to decipher the math afterwards because at least they know high level, what we’re trying to do and why.
Hellrigel:
Right. But even, [01:09:40] the Signal Processing, magazine has an article about Apollo. It would be interesting to get articles like that without the equations in them and package them for history courses. If I were still teaching my university history and technology courses this would be a prime example [01:10:00]. When I’m teaching about Apollo, even in the U.S. survey course, I can use that article.
Acero:
Yes, but I would like to have both because there are some audiences that would say well, this is all I need. But some who want everything, it is still easier to learn it this way than learning it diving right into the theory. And he’s like, oh, my God, why am I doing this— what’s the point [01:10:20], do I need to do all these things, maybe I care about other parts, other paper, not this one. But I still like to have the big picture.
Hellrigel:
Right. Right. Wade into it.
Acero:
Just get the high-level big picture. It’s easier said than done and you have masters here.
Acero:
They know how to do it, so no wonder they’re treated as rock stars.
Hellrigel:
Right [01:10:40]. Are you thinking about retiring? How do you know when you’ve had enough?
Acero:
When someone asks me this, I think why should I retire if I’m not tired. First, I need to get tired then I will retire and get tired again. But I’m not tired, I have too much energy. [01:11:00] The last time I was on a long leave was when I switched jobs and that was three months. I thought I was going to have no fingernails left. I cannot imagine retiring, I don’t know what I would do. I’m sure that if I were to retire, for the first two or three weeks it’ll be fine [01:11:20] and afterwards, I will think, so what am I doing. It’s not like I have a lot of hobbies.
Hellrigel:
Have you had any startup companies of your own?
Acero:
No.
Hellrigel:
Did that ever appeal to you?
Acero:
It has crossed my mind. The very first time I joined Apple, I don’t know if I mentioned this yesterday, but the office was on the second floor of the Cupertino [01:11:40] National Bank.
Hellrigel:
Yes, you told me about that yesterday.
Acero:
This was like the next Macintosh, the top-secret project, the most secret project that I’ve worked with really. The manager of that team would have regular staff meetings and in one of them he said, well, okay, how many people here have [01:12:00] been in a startup, raise your hand. Everybody raised their hand except for two people, Kai-Fu Lee who hired me and myself. We looked at each other and said, oh, we are the odd ones out here.
Hellrigel:
So, all the others were part of that dot-com era or some other kind of tech startups.
Acero:
And before then. That was before dot-com [01:12:20] startups. There are more startups now, but we’re talking about 1990. In 1990 and there were still lots of people that had been in some startup. Then the next question is well, how many of those startups are still alive and then a few hands. So, that was the [01:12:40] welcome to the Silicon Valley. That’s how the Silicon Valley is. I remember when we moved there and the kids, both boys were into soccer, and on teams. Mom will take them to practice during the week and dad will take them to the games during the weekends. I would go and all the dads were there. I was talking to [01:13:00] one of the dads and he was telling me, oh, I work in this company. Then a month later we met at another soccer match and so I said, how is that company doing, oh, no, no, I left that company. I’m at this other startup. I said, okay. Then a few months later, same thing, so that’s something that I had not seen [01:13:20], the kind of mobility, workforce mobility.
Hellrigel:
Well, mobility or instability?
Acero:
Well, I think that in the Silicon Valley people view this like, I know I have my network, I have a picnic with three or four people that are in startups and I’m working for this one. If for some reason that doesn’t work out [01:13:40], I’ll work for that one because the tradition was well, I’ve a family, I have a mortgage. I need to take this seriously, right, that’s the more east coast, but in the Silicon Valley, some of these people have families, have mortgages and they say, it doesn’t really matter.
Hellrigel:
Yes, that’s a significant difference. [01:14:00] My friends ask about retirement, but I can’t see doing it. I worked since I was fourteen years old.
Acero:
Right. So now, of course, I know that I can. That answer won’t work in a number of years. I would love to be like Ron [Schafer] and Al [Oppenheim] at that age. I would like to get [01:14:20] to that age. I know that at some point, that will change, but I don’t think it’s now. Will it be next month, next year, five or ten years, I don’t know.
Hellrigel:
What do you do for fun now, hobbies?
Acero:
Again, I don’t have a lot of hobbies, I just do work. When I have a little time, sometimes I go back and play the piano [01:14:40], which I always liked. Particularly if I’ve had a tough day or something, it helps me decompress and just play for a little bit. I’m a big soccer fan, so I watch all of the games of my team, Real Madrid.
Hellrigel:
That’s what I was going to say, Real Madrid, yes.
Acero:
So [01:15:00] I watch. Fortunately, they are the winningest team in history, in competition, so it’s good to be a fan. It’s not just a fair-weather fan because my dad was a season ticket holder of Real Madrid, which now is a very coveted thing. This is a team kind of like the Green Bay Packers that are owned by their [01:15:20] members, by the season ticket holders.
Hellrigel:
A season ticket holder, so you went to every—
Acero:
No, no, my dad.
Hellrigel:
Oh, your dad.
Acero:
My dad, and then—
Hellrigel:
As a boy you went with him?
Acero:
No, because what happened is when he was dating my mom, and at some point, she said “Well, it’s Sunday, let’s go out.” [01:15:40] He said, “Well, not today, I’m going to the game, you’re dating me. What’s up.” At some point, she said, “Well, you have a choice, soccer or me.” She won, and he gave up his season ticket.
Hellrigel:
He couldn’t get tickets for a few games?
Acero:
It was before they got married, [01:16:00] so I was not there.
Hellrigel:
He had made that choice.
Acero:
Yes.
Hellrigel:
Later, maybe once he had children, he could have bought another season ticket or single game tickets?
Acero:
I don’t know. Yes.
Hellrigel:
That was a committed man there.
Acero:
That’s right.
Hellrigel:
He gave up his season ticket.
Acero:
A family man. He was definitely a family man.
Hellrigel:
But then he followed the team on radio and TV?
Acero:
Yes, yes. Then [01:16:20] of course, we were from the city. Even though he was no longer a season ticket holder, my father still liked the team, so we watched it and are fans. Now, my older son is a big Real Madrid fan, too. The younger one follows but not the same. But the older one, we all both the—watch the games [01:16:40] on TV and everything.
Hellrigel:
They watch the World Cup, too?
Acero:
Yes. Well, the older one. The younger one likes to play just as much, but he’s not as excited watching.
Hellrigel:
Yes, I await the women’s World Cup.
Acero:
Yes.
Hellrigel:
I’ve seen the U.S. women play and I’ve been to NJ/NY Gotham games.
Acero:
I didn’t watch that.
Hellrigel:
[01:17:00] I played rugby, not soccer. Real Madrid’s got a loyal fan.
Acero:
Loyal, yes.
Hellrigel:
It is nice that you kept that family tradition.
Acero:
Yes, so playing piano and watching soccer. Then I try all these wines, so we’re members of a number [01:17:20] of wineries. We’re fortunate to live in wine country.
Acero:
Next weekend is our anniversary and my wife’s birthday, so I’m taking her to Napa. We’ll spend the weekend there and then stop over and pick up our wines. [01:17:40].
Hellrigel:
Congratulations and enjoy the trip. Then it’s back to work.
Acero:
Yes, but it’s fun.
Hellrigel:
What other conferences do you attend?
Acero:
Well, ICASSP has been the main one. That’s the one that is the longest record, it’s thirty years or so. Then there’s other workshops that IEEE SPS organizes in the, the speech area. ASRU, automatic speech recognition and understanding and then SLT, more recently, spoken language technologies. And they typically are in December, and they alternate. So, I haven’t gone to all of them, but I’ve gone to a number of them.
Hellrigel:
[01:18:20] The IEEE Signal Processing Society, SPS, is your home.
Acero:
There is another society called ISCA, International Speech Communication Association. They are speech conference, a speech organization. It’s much smaller than IEEE, much smaller than Signal Processing because signal processing contains speech, image and sensors, and audio - -.
Hellrigel:
Right. That’s your other institutional affiliation?
Acero:
Right. Typically, they have their main conference called Interspeech in the fall, [in the] August, September [01:19:00], October timeframe. I’ve gone to quite a few of those, not as many as ICASSP, but quite a few of those also.
Hellrigel:
And that keeps you busy enough?
Acero:
Right. Then occasionally, I manage teams that do language processing. [01:19:20] I’ve gone to other conferences that are not either one of them. One of them is NeurIPS, which is very popular these days.
Hellrigel:
Which one?
Acero:
NeurIPS, Neural Information Processing Systems. It used to be called NIPs, but it changed names a few years ago. [01:19:40] That’s now probably the preeminent machine learning conference. Traditionally, that conference was in Whistler, British Columbia in December. They had a tradition where you have some talks in the morning [01:20:00] then you have the afternoon free to go skiing. Then you come and have dinner, and then have poster sessions in the evening. That’s the way it was. It was small, very, very small.
Hellrigel:
I heard it’s beautiful in British Columbia.
Acero:
Yes, it was, but that conference has changed dramatically. ICASSP has grown over the years. [01:20:20] That conference I haven’t gone to as many, just a few, but when I went there, to Whistler, it was a fraction of what ICASSP was in size. It was somewhat comparable to one of the bigger workshops that we have. But it just took off in 2016—no, 2017, with the craze of [01:20:40] deep learning. Registration sold out in twelve minutes.
Hellrigel:
Twelve minutes. Oh, so now I understand some of the anxiousness that was spoken about during the evening talks at ICASSP. I was trying to figure out what organizations people are joining and worrying the SPS folks. Some of the speakers were worried about a drop in membership [01:21:00] because new people entering the profession do not realize the broad scope of signal processing. They are into machine learning, and they do not associate that with the IEEE Signal Processing Society.
Acero:
Right. This is definitely one.
Hellrigel:
I know the closing ceremony is in fifteen to twenty minutes.
Artificial intelligence versus machine learning, I’m not an expert in the field, but to me are these subsets of [01:21:20] machine learning? How is that different that artificial intelligence?
Acero:
So, most people say first of all there isn’t an obvious answer, well, not one that everyone would agree, but most people tend to think that machine learning is a part of artificial intelligence. Artificial intelligence includes things that are trying to simulate [01:21:40] or mimic—not necessarily imitate, but mimic or have similar behaviors as humans, the intelligence of humans. Machine learning is one way of doing that, but historically, they were doing these by building expert systems. Essentially, they interview an expert, and they write rules that try to mimic [01:22:00] the behavior of that expert. Machine learning is a term that it was coined in the late 1950s.
But it was not popular and as the technology developed and as more data came in, say well, it’s much better than just building rules. [01:22:20] Just give me the data and the system will learn from the data, as opposed to —. So, the machine learns as opposed to you learn and then you program the code. But it is a subset. Now, you may argue that these days, AI, the subset is very big. And the majority of AI is machine learning, but not all. There are other parts [01:22:40] of AI that are not machine learning. There’s also a big overlap between machine learning and signal processing.
Hellrigel:
Right. Right. So, to me, a non-expert in the field, they seem so interrelated, but there are certain specialties. It is almost as if looking at the entire set of historians in my world [01:23:00], but then there are specialists in diplomatic versus military versus international relations. They might all do sort of similar things, say the history of World War I, but if you’re an expert, you could see the differences.
Acero:
They are. Yes, and again, we mentioned at the panel the other day that some of the key concepts were [01:23:20] proposed by people that were traditionally considered signal processing people, so the field has evolved. There’s a part of it that is more like, tinker, you have these libraries, and you stitch them together and you program them and you don’t need to know the theory to do that. So, that’s probably more like machine learning, practitioners. [01:23:40] But then the people that are building the algorithms, they share a lot in common with signal processing, and there is a big overlap. Now, there’s some parts of signal processing that are not machine learning and there are parts of machine learning that are not signal processing, but there’s significant overlap. Part of the reason there is a 50 percent [01:24:00] increase in paper submission acceptance this year, just going through the sessions, you see a lot of deep learning, left and right. So, then maybe all of that is what’s causing the growth, which is good. That means that we’re embracing it. It’s almost like when the Internet came out and there isn’t an Internet conference. [01:24:20] There are seminars and meetings and everything, but it’s not like the other communities. They say, no, I don’t want to touch the Internet, Internet is everywhere, of course I’m going to use it, I’ll look at the tools and I will use it as appropriate and should.
Hellrigel:
Right. Maybe ICASSP registration didn’t sell out [01:24:40] in twelve minutes, but applications to present a paper were high, so they added an extra day. They extended the conference from June 9th to the 10th. I don’t know if it’s going to take a while for the COVID effect to shake out. For example, students, when you’re learning virtually [01:25:00] and you’re not on campus interacting and encouraging them to join IEEE it is more challenging to keep them engaged. IEEE student membership has increased tremendously in Region 10, but less so in other Regions. When I taught at university, the joke was, give them free pizza and soda, and they will attend campus events.
Acero:
Yes, yes. Absolutely.
Hellrigel:
It has taken a year [01:25:20] or two to see the numbers rebound.
Acero:
Well, the numbers again, have increased 50 percent. The society growth has not increased anywhere as much, but it’s at least not decreasing.
Hellrigel:
Right. Right. Stability and overall retention are important. I attend the monthly Membership Development meetings and people are really pleased with the renewal or retention rate, including higher-membership grade renewals and student retention.
Acero:
I thought in the various committees [01:26:00]. I’ve always thought about what Apple does, and what we can do to satisfy the users. Well, what can we do to satisfy the professionals, the members? I think sometimes, we don’t think hard enough in IEEE about this and today if you look IEEE membership, it’s been growing steadily [01:26:20] over the years. In just a few years ago, starting to top and kind of gradually slow down. It is a tale of two or three stories. if you split into academics, students and industry you will see a very different picture. I believe industry membership started dropping [01:26:40] in 2002.
Once Xplore [IEEE Xplore] came out. A big reason why people in the industry were members of IEEE is so that they could get access to these publications. But now, many of them work for companies that have institutional access through Xplore. So, they can access the publications and they don’t need [01:27:00] to be IEEE members to access them. The main reason for them to be members disappeared. That is how (industry) membership has been decreasing steadily. The point that we have less than half of what we had at the beginning of 2000. But because academics have been continued to grow and students, too, even though the industry [01:27:20] was dropping already in 2002, you couldn’t tell if you just look at the total membership number. It was growing faster.
Hellrigel:
The overall IEEE membership number.
Acero:
IEEE overall.
Hellrigel:
All right. The overall IEEE membership number increased despite the drop in industry membership.
Acero:
I don’t have the statistics for Signal Processing [IEEE Signal Processing Society], so I’m not really sure.
Hellrigel:
It’s important to break it down.
Acero:
I’ve been asking for these numbers. It’s hard to find, you cannot [01:27:40] correlate it.
Hellrigel:
They’re there, but where do you look and are they the most up to date.
Acero:
We changed the systems. I don’t know, but that’s as much information as I could get.
Hellrigel:
This is interesting because some people, at least superficially think that academics, say why join [01:28:00] if the university has free access to IEEE Xplore. However, membership also gives you outlets to publish and conferences as a lower fee.
Acero:
Well, I think academics have another big incentive, which is the prestige. In many cases, if you want to get tenure, let’s say you need to be an IEEE Fellow for instance, or something like this. If you’re not, then you’re not going to get tenure. [01:28:20] You want to do it so that you can get tenure or be more like all your more senior faculty members. Well, you see the associate professor, you see the full professor. They’re all IEEE members, so you feel like you need to be like the Jones or else you’re going to be an outcast.
Hellrigel:
Right. Right.
Acero:
[01:28:40] It’s become so much a necessity and you don’t do it for the service that you get necessarily, you do it because that’s how you move up. You get elected to all these committees and that’s valued by the university. You become a Fellow and that’s valued by the university. But none of that matters in industry by and large.
Hellrigel:
Right. Right. Then there’s the other group that I don’t know [01:29:00] if they consider them industry or academia, but federal research labs, like NIST in the U.S. (National Institute of Standards and Technology).
Acero:
Research labs, I think they are closer to academia in that sense because probably—just like when I was in industry, Microsoft, when I was in Microsoft Research, so yes, it was industry, but in [01:29:20] practice, for all intents and purposes, we were more like academics. We all wanted to become an IEEE Fellow, too. And they wanted to get an award, and they wanted to—it’s not necessarily for the tenure, but—so that was—so all these research labs are—behave more similar, but the [01:29:40] majority of people in industry are not there so that’s a small pocket.
Hellrigel:
Then too looking around all the Ph.Ds., industry probably has fewer Ph.Ds. than academia?
Acero:
Yes, true. For instance, when I last looked at my team’s composition, a few years ago, I haven’t done this exercise since then [01:30:00] because once the team gets so big, and it’s not like I have all these data easily entered. Almost half of the people in my team had a Ph.D. Now, in academia of course, every professor has one. So, yes, the percentage is higher, but other parts of the company didn’t have anywhere near as high [a percentage]. [01:30:20]. So, at least my team was more specialized. You need that extra level of research knowledge to do the kind of things that we have to do.
Hellrigel:
Wow. We are nearing the end of this recording session. It may be the time to ask if you are content with what you have accomplished thus far? [01:30:40]
Acero:
Well, yes and no. I’m content that if I look backwards and we’re at and that I had a big opportunity to play a role together with many, many people that I worked with, that I’ve been very fortunate to work with, very smart people and they are the ones that I have to hand all of these. That we’ve [01:31:00] improved the technology and that’s very rewarding. Now, I still get reports, bug reports of all the areas where (Siri) we still don’t do a good job, means that we need to keep improving. And obviously, there are times where there’s incremental and occasionally, there’s all these [01:31:20] disruptions. in 2010, when we came up with deep neural networks that was like a step function. Then it was kind of gradually growing after that, and now you get this new technology with these generative AI. Now you can see another disruption. [01:31:40] That’s the way it is. It’s not like a straight, straight line with slope, it’s more like yes and then disruption.
Hellrigel:
Right.
Acero:
Well, now, there is all this new set of technologies, so it’s exciting to say, well, what can we do with them, right? From that sense, I’m happy that I’m in a place where there’s all of the action. The bad [01:32:00] news is that people say, well, fix this for Siri. The good news is if you’re working in other jobs no one cares or you do something new, but it is exciting, you’re in the middle of the action. I always thought when I was going to school that if I had been born earlier, I would have loved to be part of some of these disruptions [01:32:20] like the ethernet or the Internet or the first personal computer as opposed to like it was invented before now. I’ve been fortunate to go through some of these. I’m really happy that I’ve done this. So, from that point of view I think there’s still a chance I can ride [01:32:40] one or two more waves.
Hellrigel:
Right. Right. Alan Oppenheim and Ronald Schafer, they had oral histories for the IEEE Signal Processing Society’s fiftieth anniversary history project. So, like them, in years to come, you can record another oral history to cover the latter part of your career; oral history number two for you. Perhaps you’re at [01:33:00] their stage twenty-five years ago. Then you’ll be a senior’s age twenty-five years from now and signal processing technology and markets will have moved ahead or developed. In twenty-five years, this will be the 100th anniversary of the IEEE Signal Processing Society.
Acero:
Well, I hope that by then I have that opportunity.
Hellrigel:
Yes. You could just see that you’re in [01:33:20] their wheelhouse and that must be a sense of accomplishment.
Acero:
Plus, I’ve always enjoyed history. When we wrote our book, we did something that we didn’t see in very many books. We had the regular chapter and then we had references and then we had a section that was a historical perspective [01:33:40]. We added references along the way, but it’s more like now, let me tell you a little bit of our view of history. Not history, but our view of history because we can’t claim that we did enough homework. It is also how we understood what happened.
Hellrigel:
Your [01:34:00] snapshot of history from your perspective.
Acero:
Yes, the way we saw history. A lot of what we do is predicting the next sample, the next thing. That’s what the technology does. Well, we should use that in predicting our ideas. Let’s look at the past, how things evolve over time, and maybe there’s a few lessons [01:34:20] that can tell you things that appear, and that appeared to be useless before, too.
I think I mentioned the other day about the CELP coder, Bishnu Atal. That’s one example, but when I went to Microsoft Research, one of the first projects that they had was called Tiger. [01:34:40] Why Tiger? It was a big server with lots of hard drives that were serving video, so that you could go to your terminal and watch a video through the terminal. Now, this is okay, not a big deal. You can do it on your iPhone. You can do it on TV with Netflix. What’s the big deal? Well, we’re talking about 1993. [01:35:00] At that point, you just take your VHS tape. This is before the DVDs. Why would you want to do this; it’s expensive. It was Tiger because to get the data, you couldn’t do it from one disk, it was too slow. You needed to stripe it, [01:35:20] and get it from multiple disks at the same time because otherwise you will not be able to feed (it in time). It was incredible. I thought why are we doing this? This is so stupid. It’s much faster to stick the VHS tape, it is much cheaper and much higher quality. Why [01:35:40] are we doing this? Well, I learned my lesson. You can’t think as Doc will say to McFly, Marty, you are not thinking fourth-dimensionally, you need to think about not where the puck is now but, where it will be.
Hellrigel:
Anticipate.
Acero:
Yes, and this is clunky. Yes, [01:36:00] you can see that that will improve, right, so if that was not clunky and was expensive, wasn’t slow, what do you think of just going to your phone and just doing—this is supposed to stick the tape, where you need to find it and—no—yes. So learn to think that way. It’s something that you get if you’ve learned a little bit of history. History of technology [01:36:20], where you can say, well, I’ve seen in half of those cases that now, I look at new thing and I just don’t think it’s totally crazy.
Hellrigel:
Have you ever worked on a project that bombed?
Acero:
Yes, of course. You need to look there. We look at machine learning as classifiers. [01:36:40] This is the class that was good, and we should invest, this is class that wasn’t. Well, when the algorithm converges, it’s easy to see which one is the binary, or the negative. something is not working, is not working because it’s a bad idea or [01:37:00] is it that is not working—it’s a good idea, but you have a bug, and you need to find it. Or maybe you don’t have a bug, but you don’t have enough data or why is it. The answer is when you replicate it with somebody else’s own then you know. Because they publish a paper, they got these results, if I don’t get them, I have a problem, I need to fix it. But if you do research and no one has published this, well, haven’t they published it because they tried it and it didn’t work and because it didn’t work, of course, they couldn’t write a paper. Or you are the first one and you need to get it to work. It’s not working now, but it may or no, no, no, it will never work. You don’t know. You don’t know. So, you need to know when to stop. It is easy [01:37:40] to start projects, but it’s not easy to sunset them. But sometimes, if you want to try new ideas, you have to sunset some. You cannot have infinite. The exponential doesn’t continue, you have a finite number of people.
Hellrigel:
Right. Yes. Well, the publications and conference presentations make research public. [01:38:00] It’s always nice to do something new, but then when do you sunset [retire or end] it?
Acero:
Yes. It’s something that you learn. Again, that’s why the history of technology is useful and there are the journals of negative results, too, where you learn this thing looked reasonable too, and how can you tell when it’s time to say, well, I gave it a fair shot? [01:38:20]
Neural networks were very popular in speech in the late 1980s and early 1990s. They were in hibernation for basically, almost twenty years until they came back. It wasn’t that it was a bad idea, it was certainly a good idea, but the planets were not aligned yet. There was not enough data, not [01:38:40] enough to compute. That may be the reason. Sometimes it is good to say well, this idea didn’t work, I’m not going to bury it, and put a flag there. I’m going to put it in my drawer because who knows. It’s in my mental drawer. Maybe years later, things change, and you say, oh, maybe that is [01:39:00] the reason that that thing didn’t work.
Hellrigel:
Right. I went to an exhibit about the videophone at the Tandon School of Engineering at New York University. [Ed Eckert, the archivist at Nokia Bell Labs also attended] and mentioned the impact of monopoly rights and other regulations on the commercial development and introduction of the videophone. You had to look at the business, the technology, the monopoly franchise, and utility regulations. [01:39:20] So, it wasn’t just the technology, it was legal infrastructure, consumer demand, and the cost of things.
Acero:
Yes.
Hellrigel:
It’s often complex.
Acero:
There are many reasons. But that’s why I developed an interest. We all did, we wrote this part of history.
Hellrigel:
But I have to look into that.
Acero:
It’s useful [01:39:40] to just study enough, so that you can see what these people did, because it can help you. I’ve seen this situation before, what should I do. Now, you have some data. It’s not that you need to make a decision without any data. Now, you don’t have all the data, but you have some data, better than someone else that is just new [01:40:00]. And you have your instincts that have been trained from all this history.
Hellrigel:
Yes. At the History Center, in our library we have a collection of textbooks. I don’t know if we have yours, but we have a number that have been donated and come in. [01:40:20] We collect them because usually libraries, when the new textbook comes in, they throw the old one out.
Acero:
Yes, because no one buys it anymore today.
Hellrigel:
Right. We don’t have room to collect everything. But we think as the History Center, many of our members have published the books. We can’t collect all their books, but [01:40:40] we’ve been lucky enough to collect a number of them.
Acero:
Well, at least the ones that were more popular at the time.
Hellrigel:
Right, right, right.
Acero:
Because that’s more representative for that time, right.
Hellrigel:
Right. While I wasn’t fangirling over Oppenheim and Schafer, it gives me context to the [technical] books that we have. You know, who were these people and [01:41:00] who authored the well-known books in various IEEE fields of interest? We [the IEEE History Center’s research library and technical book collection] have Fred Terman’s book and other early publications. It’s nice to meet the people. And is there anything we did not cover?
Acero:
No, we covered a lot of ground. [Laughter]
Hellrigel:
So, then I will let you go, sir. [01:41:20] sir.
Acero:
Okay, well—
Hellrigel:
Thank you.
Acero:
My pleasure.
Hellrigel:
Thank you very much and thank you for spending a two-parts days. If you need the facilities before you leave?
Acero:
Oh, I’m going and stretching.
Hellrigel:
Okay.
Acero:
When do you think, roughly, and I know that it’s not going to be soon, that this would be somewhere on some website or where I can take a look? [01:41:40]
Hellrigel:
Oh. Well, a minimum.
Acero:
I know that you’re going to have a backlog now, so I understand.
Hellrigel:
Right. It will be a minimum of six months before I send you the transcript for your review. I am recording five oral histories at this conference for the IEEE Signal Processing Society, and I am working on many other oral history projects, too.
Acero:
Okay. So, a few months.
Hellrigel:
Yes. Yes, a minimum of six months pending the intervention of additional projects. It is just me working on oral histories and sometimes an unexpected project or need crops up and I am reassigned to a different task.
Acero:
All right.
Hellrigel:
Yes. We are trying to get things up and running as well as processed to celebrate the IEEE Signal Processing Society’s 75th anniversary as quickly as possible. Thank you [01:42:00].
Total duration: 04:48:00
