He's been an ASU graduation reader for decades. Now an AI system uses his voice to say the names

Graduation season is almost over, which means thousands and thousands of names have been read aloud as graduates walk across stages to get their diplomas. But new technology has led to some challenges and controversies.

Officials at Glendale Community College this week had to apologize after the AI system they used to read graduates’ names malfunctioned.

By way of full disclosure, KJZZ is licensed to the Maricopa Community College district, which also includes Glendale Community College.

Peter Lafford, associate research professional emeritus at Arizona State University, joined The Show to talk more about the evolving role of the graduation name reader — he’s also read names at Arizona State University graduations for three decades.

Peter Lafford

Full conversation

MARK BRODIE: Peter, good morning. Thanks for being here.

PETER LAFFORD: Good morning.

MARK BRODIE: So, what is like, you, you still do this? So, what has your experience been with using AI?

PETER LAFFORD: Well, it’s, it’s been great. First of all, it isn’t Siri or Alexa, or E.T. if any of your audience remembers, "Hi, little voice, I’m E.T." No, no, it’s not that, and, and we’re not doing deepfakes.

It’s that we’ve created AI voices of my voice reading names. The company doing this for us, Tassl, and ElevenLabs with the AI voice, had us read like like an hour’s worth of names where they’ve got the actual name we’re reading and the sound we’re making and they use that to create an AI voice that I put my own words into and it comes out just fine and perfect.

I don’t know, do you have do you have that sample I was able to give to you of what it what I was doing at ASU last week?

MARK BRODIE: We do, yeah. Let’s let’s take a listen to the the AI version of Peter Lafford here.

PETER LAFFORD (AI VOICE): Abdul Qayyum Abdul Rashid Khan. Srupthi Hosahalli Manjunatha. Yuva Vignesh Balamurugan.

MARK BRODIE: So that was, Peter, that was you as AI reading the names of graduates.

PETER LAFFORD: Exactly, exactly. And and in fact, the company takes the names that they want to be announced, submitted to the AI voice, give the students a chance to go, "Thumbs up, yeah, that’s good," or "No, no, no, try again, try again."

And if they still don’t like what the AI voice was doing for them, it kicks over into the reader record queue where we, the readers, actually record it in our own studio, and this is a couple of weeks, you know, before the event actually.

MARK BRODIE: Yeah.

PETER LAFFORD: And so we can record it then because they didn’t like it. So all of the students have a chance to say, "Yeah, that’s good." And that’s why it it’s indistinguishable from my regular reading voice, what the AI voice is doing. It’s really quite quite impressive.

MARK BRODIE: Well, and this is of course, the main goal, right? Like is to get the students’ names right. They’ve worked hard, they’ve achieved something, you don’t want to screw up their name on what one of the biggest days of their lives.

PETER LAFFORD: Exactly, exactly. And I mean, it evolved from when I was just reading the names that had been scribbled on an index card and held in their sweaty little palm for a half hour, I had to unroll that, quickly see the name and read it live.

Now they get onto the webpage, they can even put annotations of what they think it is, they can even record their own pronunciation that we we get to hear and listen and we can make annotations or as we now record the names in advance, we can tweak it to see to be just what they recorded for us, and now we can let AI do it and even tweak it more if they didn’t like what AI was doing.

MARK BRODIE: Do you think that there is anything lost when you don’t have a human on a stage reading a graduate’s name?

PETER LAFFORD: I, I don’t. I mean, I think, it’s been fun that when I run into somebody who says, "Oh, I graduated from ASU," I said, "What school? I might have read your name at graduation." Well, now I’d be saying, "Well, oh, oh, you went to ASU. What school? Oh, yeah, your name might have been read by my AI voice."

But I trained that AI voice, AI voice with, you know, years of experience in languages and I taught English as a second language, so I encountered students from all around the world. I’ve helped make it work that well that I do feel good about it that they get it get it right.

MARK BRODIE: And obviously, this is a very new technology, so not a huge sort of sample size here, but it sounds like you’ve not had the experience that folks at Glendale Community College had earlier this week when the system, the system just didn’t work as planned.

PETER LAFFORD: Right, that wasn’t an AI problem, that was an AV problem. The computer glitched or their audio connection or their scanners weren’t working. Something strange was happening there, but the the names that it was playing, you heard, they were just fine and dandy. It’s just that the system wasn’t working right. They didn’t have a backup plan that worked quite right, so that’s where the the trouble came in.

MARK BRODIE: So is the key then to have a backup plan where like, yes, let’s use the AI if that’s the system we’re set up to use, but there should also be a live human there who’s prepared to to read names if need be?

PETER LAFFORD: So, there, the the way that Tassl recommends people, schools do it, is to have two computers side-by-side, you know, the one they’re using with the scanner they’re using, and then a hot swap right next door ready to plug in if something were to go wrong with that. And then you still have the person at the stage who is taking the card and scanning it. And then if it doesn’t work, then they get to read it.

I mean, funny thing happened back after COVID, when we were finally back in events at ASU. We were all wearing masks and I had prerecorded all of the names, and but I had my mask on and I scanned the first name and the name displayed on the screen but nothing played on the audio. I scanned the next card, nothing on the audio, phooey.

And so I went ahead and read it, read it live, went through the rest of the ceremony reading the names live. But when I went to talk to Melissa Goitia, who is the coordinator there, I said, "Hey, what happened?" She said, "Oh, that went well." I said, "No, no, I was reading all of the names."

MARK BRODIE: She had no idea.

PETER LAFFORD: The fact is, the mask, I thought, the mask was going to hide the fact that I wasn’t saying the names live. In fact, it head, hid the fact that I was, was reading the names live, so the joke was on me, you see.

But now it’s the same thing, those, those names whether they were done by my reading voice or my AI voice, they’re all on the system ready to go, and so that that shouldn’t really be be an issue.

MARK BRODIE: Interesting. All right, we’ll have to leave it there. The real Peter Lafford, he’s been reading names at ASU graduations for three decades. Peter, thanks for your insights.

PETER LAFFORD: Thank you so much, Mark.

KJZZ's The Show transcripts are created for audience accessibility. Transcripts are created on deadline with the assistance of AI tools and then edited, and may not be in their final form. The authoritative record of KJZZ's programming is the audio segment.