Auto-Tune always and forever

Welcome to the Verge Cast, the flagship podcast of the difference between Reverb and Dverb. I'm your friend, David Pierce, and I have decided that I am going to relearn how to play the guitar. So I have this guitar that sits behind me on all of my meetings and all of my podcast recordings. And so people are often like, "Oh, that's cool. Do you play guitar?" Which are favorite songs? Do you know a time of your life by Green Day? Like all this stuff. And the answer is, I don't really play. I used to. I played a lot when I was a kid for a long time. I was actually pretty good at it. But then I got to that point that you get to with a lot of things where my guitar teacher was like, "You either need to try harder and care about this more," or just quit. And so I just quit. Bad choice in retrospect, but like I was 12, what are you going to do? But now I have this guitar, and I've decided I'm going to remember how to play it. So I got this app, "Usation," that people really recommend, and I really like so far. It just sits here on the iPad, and it tells me how to play, and it actually uses the microphone to see if I'm playing it correctly and gives you this sort of dynamic feedback as you go. It's not as good as having like a person to teach me, but I can do it in my basement. So I'm calling it a victory. So far, I have learned basically that this is a C chord, and not much else, but you know, it's progress. We're doing it one day at a time. Anyway, that is not what we're here to talk about. This is the second episode in our miniseries all about the future of music. Last week, we talked to Jack Cohen about trackstar and basically music content and how we as fans discover new music and how musicians discover new fans and be in the world. This week, we're talking about a technology that I think you could argue is the single most important thing that has happened to music in the last two decades, and that thing is auto-tune. We're talking to Charlie Harding, who is a longtime Verge cast contributor and a friend of the show, and also the co-host of the excellent podcast "Switched On Pop." I'll link it in the show notes. If you don't listen to it, you should. It's excellent. He's going to tell us all about the history of auto-tune, how it changed music, and maybe try to figure out with us if there's a way to take what happened to auto-tune over the last two decades and think ahead. As we get to AI and TikTok and all the other changes that are happening to the music industry, what can we learn from what has happened with auto-tune that might give us a hint about where we're going the next 25 years? All that is coming up in just a sec. We have a lot of very fun songs to play. They are going to stick in your brain for months, and I'm very, very sorry about that, but it's going to be worth it. It's going to be awesome. All that is coming up, but I just remember how to play a D-Cort, and so I'm going to go do that a bunch of times and feel very good about myself. I know two chords now, and basically a guitar player. This is the Verge cast. We'll be right back. You know that feeling. Too many updates, too many meetings, too many things, falling through the cracks. Monday.com was built for that gap. The AI work platform designed from the ground up for people and agents to work side-by-side and deliver more together, drafting the brief, flagging risks, running a competitive scan, your agent can do it all, and the best part, anyone can build one. Even you, try it, create your first Monday agent today at Monday.com. Welcome back. Alright, let's just get into it. So Charlie Harding, like I said, is a music journalist, he's a professor of music. He's also the co-host of Switched On Pop, one of my absolute favorite, deep, wonky music podcasts. I really cannot recommend it enough. He's been on this show with us before, talking mostly about AI music. We did a great episode of him last fall, where we tried to make a song with AI, and it got weird and very fun. He was around for the whole laser-bong craziness. We talked a lot about AI Drake with him. We've talked a lot about AI with Charlie. Today we're going to talk about something different, and that is auto-tune. But one of the reasons I wanted to have Charlie here to do this is that I think there might be some similarities between the story of auto-tune and the story of AI. So in order to do that, Charlie is going to show up here, and he's going to tell us the story of auto-tune, how it changed music, how it changed the world, and then we're going to try to figure out if there's anything we can learn about what comes next. So here we go. Let's just get into it. It's my conversation with Charlie Harding. Charlie Harding. Welcome back to the show. Pleasure to be here. It's been a minute. This is very fun. I was having a reason to talk to you. And this time, it's not about weird AI videos on YouTube. This is very exciting for us. I imagine we will get to AI because that is the topic. We will almost get to AI videos on YouTube. I just realized that is where we're headed here at some point, but we're going to take a minute to get there. So I want to start, this is sort of a needlessly pedantic way to start this conversation. I've realized in prepping for this that when we talk about auto-tune, we're kind of talking about a bunch of things, but also one very specific thing. So before we get into the whole story of auto-tune, like, what is auto-tune? What do people mean when they refer to auto-tune today in 2024? I think there's two different definitions of auto-tune. The sort of formal definition is that there's a company called Antares that made an audio software tool called auto-tune that is used to help people adjust the pitch of their vocal after they have recorded it, or actually, you can even record it live into auto-tune today that is also an effect people use. It's an audio processing effect, and it is also second definition basically becomes synonymous with any kind of pitch correction. There are many companies that help us do pitch correction, and so when you say auto-tune, you could either mean this audio plugin or you could mean any form of pitch manipulation and correction. Okay. So auto-tune is like the clean ex of the space. It is both a thing, but it's also the name everybody uses to refer to everything kind of like it. Yes, exactly. Okay. Got it. So let's go all the way back to the beginning here. You mentioned auto-tune the product. Yes. Where did it come from? There are so many things in audio that have come from the military, industrial complex, aerospace, and in this case, the oil industry, because auto-tune was invented by a guy named Andy Hildebrand, who was a geologist, who worked in oil and gas, and he used techniques like the FOIA transform analysis, which is used both in geophysics and in audio processing. And he helped find oil deposits using wave seismology tools. I don't quite know the science of this stuff. I'm a music journalist, and he had this idea that he wanted to pursue his passion in music. He was a floutist, and that is one who plays the flute to be clear. And so he launched this software company in the '90s and built a tool that used some of the math and science that he had used in oil and gas to create a pitch correction tool. Somehow, the wave technology he used to identify oil fields also applied to how we can better tune our vocals. So thank you, oil and gas, for pitch correction. If there's any reason you didn't like auto-tune, I just gave you another one. Yeah, right? It's for better or for worse. I have to say, no single part of that answer was remotely close to what I expected. I love it. This is great. So how does it go from guy-makes science project to actually being used? I think brief spoiler alert, we're going to get to share here very quickly. Yeah, of course. Is share the very, very beginning of that story or was auto-tune kind of a round in the industry before share happened? It's very close to share. The tool auto-tune launches in 1997, shares believe which is really the song where people start to recognize the effect isn't until the fall of 1998. The idea of auto-tune was that it would be a sort of subtle effect to help slightly enhance a out of tune vocal, not completely out of tune. You still had to sing pretty well in tune, but it could help sort of nudge the bad moments back into tune. So there certainly were some producers that were using it before share. What share did is that she and her team took this tool and kind of abused it. They set the setting so that it pitch corrected as fast as it possibly could in such a way that it created this very strange vocal effect that they started calling the share effect. I do have to note though. There is an important recording that came out before shares believe that also had this over-the-top auto-tune effect, and it's very unfortunate that it is Kid Rock's Only God Knows Why released a few months before shares believe. Yeah, that's tough. We'll give this a share. I think it's right. Absolutely, and the thing is that song was released as an album cut, not as a single, but after shares believe they were like, "Oh, that's a cool sound, and we already did it." So they later released Only God Knows Why as a single, and so it's often heard as having occurred after shares believe it was technically released beforehand. So we've got oil and gas, we've got Kid Rock, we're just, we're hitting all the marks right now. Some tough bedfellows on the broadcast today. So it's super interesting to me that at the very beginning, like months after this thing was launched for this, a sensibly pretty straightforward purpose. Yeah, and a dime totally. It's already being used in these wild out there surprising ways. Did it happen like that from the very beginning? Was it, it seems like it got crazy before it got quiet in such a really interesting way with auto-tune? Oh, the sh- Share Effect was almost a one off. Like some people started to do similar-ish things. You can think of like Madonna's music in 2000, has a sort of pitch correction, you sound, daft punks, one more time. In 2001 actually uses a different pitch correction tool, but people start to mimic this effect for sure. It's much more common that people are using auto-tune, the proper tool. This is effect as like really gentle pitch correction. And so the share effect was kind of like, I think at first kind of a one off novelty. People were using auto-tune though. Like you can hear it very audibly on a song like Maroon 5's "She Will Be Loved" from 2002. (upbeat music) (upbeat music) Where it was supposed to be subtly used, but there's something just like a little bit wrong with the recording and people picked up on the fact that like, I don't know, there's something about that vocal, which is slightly inhuman, I can't identify it 'cause it's not that like hard tuning share effect. It's the auto-tune tool, but kind of overused. And people started to complain, oh this is the sign of the end of music. People aren't really singing, they're doing something after the fact to, you know, make a glossy and shiny. Fascinating. Was that the reaction to share even because she was not trying to hide what was going on there? - No, no, well, when you listen to the song believe, it uses both a mix of tuning and no tuning. In the verse, she begins with auto-tune on. And she's sort of singing about, you know, trying to find love after a broken heart. And there's this sort of feeling of like this, you know, this robotic, just going through the motions kind of person and the vocal tuning almost matches the presentation of the lyric. Then as she goes into the chorus, she sings, "I believe in life after love." She actually drops the auto tune. ♪ I can feel something ♪ - So it's actually this creative effect in what she's saying, like, I can feel a robotic and like, soulless and then I can sort of break through and shares of course an amazing vocalist. And so she turns that effect on and off to match the lyric of the song. When it came out though, everyone's like, whoa, what is she doing? Not like, oh, she's not singing, but that is a very strange and cool effect. I want to do that too. And Sharon, her producers actually misled people and didn't tell them what they did because they wanted to have this kind of cool proprietary sound that nobody else could mimic. - That's so interesting. And at that point, auto tune I guess was new enough that we hadn't already sort of explored the edges of it in that same way. - Oh yeah, absolutely not. This was totally novel to people. They'd never heard a thing quite like it. But there is a whole history of other vocal processing techniques, tools like the vocoder or the talk box. And people probably assumed, and actually the producers misled interviewers and said, oh yeah, it's just like a vocoding tool, which is a tool that actually goes back to World War II. It was invented as a way of encoding and decoding messages to be sent under C, so they wouldn't be copied German U-boats. I'm telling you, wow, audio is like all military industrial stuff. We wouldn't have recorded music if it weren't for early military audio tools. But in any case, they said that it was this tool that had been around for decades and decades and decades so that they could have their secret sauce. So real quick, just give me like the technical explanation of how auto tune actually works. Like you mentioned, it's job ostensibly is to take a pitch that you sing incorrectly and make it the correct pitch. I understand that much. - Yeah. - What else is useful to understand about how auto tune actually works in order to understand how people use it? - Basically, it's a tool that is trying to identify what pitch are you attempting to sing. And then it does what you would call pitch quantization, basically pushing the pitch that you are singing to the pitch that you're trying to sing. And you can do this at different speeds. So if I'm singing here, but I'm trying to sing here, it would go, it would sort of glide you to your pitch. And when we think of the share effect or the auto tune effect, that is turning that pitch speed all the way down to zero, meaning it automatically immediately jumps your voice from one pitch to another. And when it does that, it creates this sort of digital artifacting that is unnatural, it is strange, and it is now very desirable. People like this sound and intentionally bake it into how they sing because it is a fun creative effect. - Got it, okay. But in theory, it really is just that simple of like, you have a pitch, and you have a pitch that you want, and auto tune makes those the same thing. - Yes, and you can say, how fast does it do it? You can maybe add some vibrato. There's all these little fine tune parameters, but basically fix my pitch. - Okay, and you do this by sort of saying, I'm trying to sing in this scale, and you set your scale, and then it just knows that, hey, you're singing way off your scale. Let me fix that for you. - Okay, so in sort of that basic definition of it, I can actually understand why auto tune, if it worked at the beginning, and it sounds like it did, would be an immediate gigantic hit, right? And from the way you describe it, it sounds like it was. You talk about Madonna, you talk about Maroon five, you talk about, like, these are A plus list artists who are using auto tune basically immediately. Did this thing just come out and take over the music industry? - I mean, it definitely today is used on the majority of recordings. And when I say majority, I mean like, over 90% of recordings. The only recordings that won't use auto tune are probably going to be, and I should say use some kind of pitch correction. The other recordings that don't use pitch correction would be like a very naturalistic rock song, maybe an indie song. You're certainly certain rappers who are not singing, use it, but the majority of any sung vocal is using auto tune. When it came out the gate, sure, I mean, any time that somebody needed to post-process, fix something that was at a tune, auto tune was the tool to use. There have been tools for quantizing rhythm that had existed before auto tune, so that you could say, "Hey, you meant to hit the snare here, but you actually hit it a little bit late. Let's nudge it back over." And so producers have always wanted to find ways to fix pitch problems. And they actually had really challenging slow ways of fixing pitch all the way back into the '70s, where you would go into a piece of tape. There were digital processes you would send it to, and you could nudge one little note a little bit north or south of where it's supposed to land. It was very slow and laborious. So people have always trying to perfect recordings that are slightly imperfect to make them more exactly in key. - So I keep coming back to the share thing, because I'm so fascinated by the idea of like, a lot of what you just described are sort of tools in a producer's toolkit, right? - Yes. - Things that in theory aren't sort of front and center in the recording, they're just designed to like solve a set of problems that you have. And I think there was clearly a lot of that in how auto tune is used and still is. And I think to your point, not every song sounds like belief, but if every song has auto tune, it's being used in the way it's supposed to. But also it became an instrument in a very real way that most of these technological things don't. - Right, and that's thanks to T-Pain, right? So it goes from being the share effect to becoming the T-Pain effect. T-Pain who likes share, great singer, which no one knew forever, by the way. My favorite thing about T-Pain, like everybody knew Cher could sing before she did believe, but we spent like a decade being like a T-Pain can't sing. And it's all because of auto tune. - Right, right. So Cher puts up belief in 1998. Fast forward to 2005, T-Pains I'm Sprung is one of the first time that people hear this effect in the world of R&B. (upbeat music) - And it goes from being the share effect to being the T-Pain effect. And pretty quickly, it becomes this sort of way that people sing. In particular, it gives rappers the capacity of singing. There are a number of early recordings, but the one that really sort of lands it for folks is Kanye's 808's in heartbreak, the song Heartless in 2008. ♪ In the night I hear a talk of the cold story ♪ ♪ Yeah, the tone ♪ ♪ Tell where far along this road ♪ - All of a sudden Kanye, who he's not a particularly strong singer, has the capacity to hit a pitch and sing melodies. Once that track is out the gate, every rapper realizes that they have the opportunity to start embedding hooks into their music. And oftentimes it is the chorus hook that people sing. And so I would say it's from Cher to T-Pain to Kanye and then everybody. That's really, it's really in those sort of late odds that the auto-tune effect becomes ubiquitous. It takes more than a decade from Cher's belief before people start using it as in the same way they might use distortion or reverb. It just becomes another tool in the toolbox that people are using as this overblown effect, not the gentle version of it, right? But I guess one key difference is that at least to my knowledge, and you're much more in this than I am, but I don't remember there being a huge cultural backlash against reverb. And yet, there have been at least a couple that I can remember of sort of big, loud, angry moments where the music industry or someone big in the music industry said, "You know, we have to get rid of auto-tune. It's ruining everything. It is destroying this thing we believe in." What just before we get into audit, make the case against auto-tune. Like if you were, put yourself in the position, you, Charlie Harding, hate auto-tune and think it should be excised from the earth. - Coming back. - I guess you could say that auto-tune homogenizes the voice. I'm having trouble saying this 'cause I actually don't believe it, but like we're just gonna clip that part and just play it. - Okay, perfect. - That's the whole thing. - Like someone's unique vocal identity is the way that they hit a pitch. Not what he sings perfectly in pitch. We all scoop into our notes. We use vibrato differently. And auto-tune sort of, you know, carves out all of those imperfections. Part of what makes us sound like ourselves makes us sound human. And so you could say that, yeah, it's a homogenizing tool. You could also say that, you know, prince stars who don't have proper vocal training and technique, but you know, that's like the history of rock and roll are people who were not trained singers who sing properly with all their air and you know, but like rather have personality to their voice and maybe intentionally don't sing. Like Henry Rollins is a great example. Like the entire word of like punk music is not about singing properly. So I think both of these arguments of it homogenizes the voice and it allows people who don't sing well to be stars are probably the primary criticisms. In the Aughts, after Kanye's record, as more and more people start to copy the sound, there were plenty of critics who are just like, everyone starting to sound the same. That's the age old criticism of all popular music. You know, everyone sounds the same. There are still plenty of, you know, boomer YouTubers who like to talk about how auto-tune is still ruining music. And the reality is that auto-tune only becomes more popular after that era listeners want more of it. It's not some cabal of music executives being like, "Oh, we are going to give auto-tune to the people." But rather it's like clearly people are listening to this music and they want more of it. The counter-argument to auto-tune homogenizing voices is that like if you listen to Drake's passion fruit, ♪ Passionate from miles away ♪ ♪ Passive with the things you say ♪ It does not sound like Charlie XC X is 360. ♪ Yeah, 360 ♪ ♪ And you're in a variety of what you see ♪ ♪ You're in a variety of just looking at me ♪ ♪ I know for what I'm so Julia ♪ It does not sound like Travis Scott's sicko mode. ♪ Make this deal with all the heists on in the booth ♪ ♪ At the gate outside ♪ ♪ When they pull up, they give me loose gear ♪ ♪ Jump up ♪ Like you can still identify who people are. And I love that about auto-tune. Is it actually, it's what you do with it. It makes it your own way of recording. I think about it kind of like. You know, when you pick up an old-style phone, there's all kinds of ways that the voice is being processed. All the low-end information and the high-end information is being cut off and all the dynamic volume changes are also being squashed and compressed and they're doing that to save data on networks. And yet like, I mean, if you call me on the phone, I would immediately know it's your voice and not Neili's voice, right? Like the ear is so meldable to whatever effects that we use to change it. I think that's awesome. So I think where auto-tune goes from here is that people learn to use this effect to enhance their own creativity, their own artistry, artists like Drake start embedding really great hooks in his wraps because he can now sing in ways that he couldn't otherwise. He's not a great vocalist. He's a great rapper and he's a great melodies. But like, again, he's not trained to be a great vocalist. Singer, you know, we don't get Travis Scott without auto-tune, you don't get Charlie XCX without auto-tune. People end up using this tune this sound very creatively and it explodes. Okay. And do you think of it in your own world and work as sort of spiritually different from any other tool that in a toolkit that a producer or an artist has like, is it, does it belong on the same level as reverb and distortion or is it something else entirely? I think about it as a producer and as an educator and music as just another tool in the toolbox for sure. I think that a lot of the backlash has to do with this question of like what is human and what is inhuman. The synthesizer had the same kind of backlash in its development that, oh, it's not a real person playing a real instrument. This is mechanical, it doesn't sound as expressive. Let me clearly, those critics also lost the plot because the synthesizer is one of the predominant sounds of popular music. I think about it as just a normal effect that anybody can use and one that should be used creatively to enhance the feeling of a song. Okay. How has auto-tune changed? It's what, 27 years old now? Yeah. Is it functionally still the same piece of software? Was it in the '90s or has it? Sure, sure. There are many auto-tune is a software developer. That's they need to constantly shout new things. They got to show poorly gross, man. Yeah, you got to keep buying the next version. So I think, I don't know, or well pass auto-tune X. I can't remember what version we're on. Their engagement metrics are off the charts. There's like a social network in there somewhere. Oh, they have subscription service. They've got everything. The reality is that actually a lot of people still like the auto-tune effect of yesterday year. And so if you use auto-tune today and you buy the latest version of it, you can turn on classic mode so you can actually go back to how it used to sound. Auto-tune today is much more naturalistic. It does a better job of tuning your vocal in ways that are less obvious. But the auto-tune effect that people love, whether it's with T-Pain or Cher or Kanye, whoever, what you like is that hard tuning the thing which is inhuman and you like to play with it. And so most people actually will just go back to the classic mode. All right. I want to talk more about where we are right now with auto-tune, but first we have to take a quick break. We'll be right back. If you're looking to hire, you need indeed. With indeed sponsored jobs, you can spend less time searching and more time actually interviewing candidates who check all your boxes. Less stress, less time, more results. And listeners of this show will get a $75 sponsored job credit to help your job get the premium status it deserves at indeed.com/podcast. Just go to indeed.com/podcast right now and support our show by saying you heard about indeed on this podcast, indeed.com/podcast. Terms and conditions applied. Need to hire? This is a job for indeed sponsored jobs. All right. We're back. Charlie Harding is still here. Hi, Charlie. Hello. So I want to talk about auto-tune kind of right now, because I think you've mentioned this a little bit a couple of times, but I think the idea that the existence of auto-tune has a changed the kinds of music that people are able to make who didn't have certain kinds of skills or whatever before, but also the kinds of people who might be coming into the music industry has changed. So I'm curious if you look at the sweep of who is making music and the music that they're making. Can you look at it as auto-tune changed all of this? The fact that auto-tune is everywhere and can make anyone singing sound pretty good. Has that changed the entire music universe? I kind of think about it more like the development of the electric guitar. It was like a new instrument that people could pick up and it created this new sound. And so I think young producers today want to sound like auto-tune. Just as in the '60s, they wanted to sound like Hendrix, right? Like it's the sound that is attractive. And so yeah, it does let anybody sing, but only let people sing if they can't sing with the auto-tune effect. And that's probably the thing that they're going for anyway. It's slightly surprising, by the way, that that sound has been popular for this long, right? Like we're a long way into this world. And again, you go all the way back to T-Pain in the early arts and share even before that. And the idea that people like the sound of auto-tune has lasted a really long time. - That's why it's more like a guitar or a distortion effect. It's a feeling that it gives people it's not about vocal tuning. Like this is what gets people angry. It's like, that makes you any human, but no, no, no, no. It's like the electric guitar is also unreal, right? Like it's generating electrical signals that has to be amplified by an amplifier. It doesn't have all the acoustic properties of a beautiful acoustic guitar, but it also sings and speaks in its own unique way. And so that's how I think about it. I think that young people wanting to produce music want to sound like auto-tune. I mean, truly, when you record with auto-tune today, it's not a post-processing technique. I mentioned it earlier. People put auto-tune on their vocal from the get-go, and they're trying to play with the effect to get it to do this weird artifacting thing. They like all the things that it does imprecisely. That's why they use the classic mode. They wanted to sound like this thing from 1998. They don't want it to sound like a pristine pure. or a tool that allows them to sing better. It's not about that. - Interesting. So explain that to me a little bit, because I think one thing I've read a bunch is that one big shift was when auto-tune went from a thing you did in post after you laid down a vocal track. Like it was a post-processing tool to now something that you do live in the beginning. And there was a line in the, there's a really good pitchfork story that you sent me about the history of auto-tune. - Oh, by Simon Reynolds, yeah, it's spectacular. There's a line in there that's basically like, there are a lot of singers who have never heard their takes without auto-tune. That is like just my brain exploded thinking about that. And so, you mentioned like thinking differently about the way that they're actually making this stuff in real time because of auto-tune. What does that look like? Like what would it be like in the studio to record in a sort of immediately post auto-tune universe? - I mean, people do this not just in studios, but in home bedrooms. I mean, the biggest stars in the world would intentionally travel to Hawaii to record and record in an Airbnb. They bring a microphone. They bring a portable small recording interface. Oftentimes, they'll use one called the universal audio Apollo that allows you to run auto-tune on this little interface live without any sort of latency so that when you're speaking into your microphone, you just hear yourself through auto-tune. And then they're singing, trying to get it to do things that make weird auto-tune sound. - I'm bad at singing with auto-tune. You actually have to be good at using the effect to make it sound right. You have to intentionally sing slightly out of tune to get it to sort of bend back into tune. You have to-- - So if you sing well, it actually doesn't get to do anything so it doesn't-- - Yeah, it's not gonna be interesting. - Yeah. - I don't sing well, by the way. I'm not a vocalist, but I am not good at using auto-tune to make it sound fun and creative. You actually have to be talented at using auto-tune and practice with it. So yeah, it's something you bake into the sound before it even goes into your software. Like you can't turn it off after the fact now. Some people still do, but like a lot of people don't wanna hear what their voice sounds like without it on. This is what I'm saying where it's like, I think of it more like the electric guitar because to play electric guitar with a lot of distortion is fundamentally different than playing classical guitar. And if you ask a classical guitarist to play a super distorted electric guitar, they might not be able to control all the feedback that it creates. They might not be aware of how to get it to sing properly and vice versa, you know, like a fast lead guitar as an electric guitar might sound absolutely terrible on an acoustic guitar. It's a skill that you have to learn to sing into auto-tune. - So you actually think of singing into auto-tune and singing without auto-tune as two just utterly different skills. - I think that they are related skills, but to make auto-tune sound good, I think you gotta be good at it. Again, it doesn't necessarily mean you have to, I'm not arguing that you have to go to a music school to learn how to do this, but you have to practice at it. Like it takes time. It's a skill that you have to develop as a musician. And when I do it, I don't sound good. It is common that people will use auto-tune to practice melodies and, you know, use it as a way of dealing with the fact that they aren't good vocalists. But again, I think that if you want to sound like T-Pain, you have to practice singing like T-Pain. - Yeah. - And again, T-Pain is a great vocalist. Share is a great vocalist. - There's just not doing the same thing as, you know, Frank Sinatra was a decade ago. - It's just, yeah, it's just a different way of singing. - Yeah. I feel like there are gonna be some people who hate that, but I find that kind of fascinating. I love it. - I think there are a lot of people that are not gonna like what I just said, but I really believe it. - We'll put Charlie's email in the show notes. You can tell it to yourself. - Thank you. I just, I mean, I just know this through practice of working with musicians in studios, and some people are really good at using auto-tune. Other people aren't. - Yeah. What, how does that pertain to live music? Is it the same, is it the same sort of thing? Like as auto-tune has gotten more powerful, it's just present live the same way it is in recordings and it's fine. - Once again, there are both ways of using auto-tune live. Some people need to have that effect on when they sing, because like if you're Travis Scott and you're performing live in concert, that is your sound. So they developed, Ontario's developed, an auto-tune live tool that you can run, you know, through your microphone on a live stage. It's not baked into the microphone. It's an effect down the audio chain at some point. - Sure. - There is also auto-tune live used for that subtle vocal tuning to help people. You know, when you're running around on stage, it's exhausting, you might be out of breath, you're not hitting your pitches as effectively. And there's all kinds of ways that vocals are enhanced to sound more like the original recording. You know, subtle auto-tuning is one version of that. Also playing backing tracks of like a thousand chorus vocals that are perfectly already in tune. It's another way that things sound more in tune when you're at a live recording, yeah. So you sing the one, but there's 50 other use singing it correctly around. - Yeah, yeah, exactly. - I want that, and that's how I sing. - That sounds great. - I know, totally. It's like, I always want my backing tracks. That's because people expect the recordings that they hear. And when you listen to your favorite artist, they're usually in the chorus singing at least nine versions of the track to add depth with chorusing. Like the thing that just makes it sound big and awesome. And, you know, bringing eight extra backup singers on tour is expensive. So we might just play the backing track and add a little tuning to your vocal 'cause you're out of breath. So those are all different ways that people use auto-tuned live, either the auto-tuned effect or the sort of more subtle pitch correction version of it. - What's the simplest, most mainstream version of that at this point? I think about the, maybe video is a good metaphor, right? Where at the very high end you have super high end video editing software, but we've boiled it all the way down to like, you can edit pretty successfully inside of TikTok. And it has made it available to tons of people who are doing tons of different things. Do we have that equivalent for auto-tuned? Like is that just available to regular people in the way that like Instagram filters are? - Yeah, I mean, there are iPhone apps that are like the auto-tuned effect that have been developed by the Gregory Brothers, who famously made auto-tuned the news years ago. ♪ You don't have a cup in confess ♪ ♪ We're looking for you, we gon' find you ♪ ♪ We gon' find you, so you can run and tell that ♪ With you buy Apple's Logic, which is their like audio recording software, it comes with their own free version of pitch correction. Every software developer that makes audio tools have tried to make their own auto-tuned effect and they are cheaper than the untaraged version. So there's all kinds of way of getting into auto-tuned if you wanna try to auto-tune your vocals. It's very accessible. - Okay, and is that a good thing? I feel like you would argue that's a good thing, yeah. - Yeah, I don't know. Or are you trying to argue that there should be like fewer brushes and paints in the world so that we can have more moneys? I think more creative tools for more people is great. - Yeah, I support it. So there is one sort of new backlash to this that I've been thinking about a lot and I'm curious to get your feedback on. So one thing I see on TikTok and Instagram and everywhere else these days is like these videos of singers in their kitchen or in like a garage or just wandering down the street, singing these beautiful soulful melodies. Do you know what I'm talking about? There's this one that's like a guy with a salt and pepper beard and he's just pouring coffee and his girlfriend is standing behind him and he just starts singing into an air pod and it sounds incredible. ♪ You let me clear up from the world ♪ ♪ You get in our way ♪ And I'm like, first of all the way I know this isn't real is because air pods don't sound like that in your mind. But like all that aside, there's been this really fascinating thing where a bunch of singers have come up and gotten a lot of fans and a lot of fame and record deals and some stuff out of these social videos. It's a very natural musician, creator way to come up. But then in every single one of their videos there is this giant comment section saying, "Oh, they're using auto-tune." I'd love to hear what it sounds like without auto-tune. It's auto-tune. And it's like on the one hand, maybe the argument is who cares. But on the other hand, it's like, we're in this moment where authenticity is everything and also you're standing in your kitchen singing. So the idea that I'm supposed to process this like a professional recording, there's a disconnect there that I have always struggled to wrap my head around how I'm supposed to feel about this. And I'm curious like if they were playing an electric guitar I'd be able to see the electric guitar, right? So I'm curious how you think about how auto-tune is supposed to fit in that version of music that we're in right now. - Well, I think auto-tune has become a sort of like catch all for any sound that is post-processing that makes things sound more perfect than they are. So for example, like if I'm singing my kitchen there's gonna be a bunch of reverb and it's gonna sound really bad. But there's all kinds of tools you can use to get rid of reverb after the fact. There are ways that you can use just EQ and compression and other very common tools and post-processing to make your voice sound better. So even if I record with an iPhone I can get my iPhone to sound pretty close to a professional recording by having good post-processing tools. And so I think auto-tune has just been, yeah, is the catch all enemy for anyone who doesn't like music which is slightly unnaturalistic. I just wanna say if you don't like the sound of auto-tune that's fine. Like, I think it's totally fine to have your taste if you prefer house synotrosings, that's cool. Like, if you prefer how punk rockers sing, that's also fine. There are different aesthetics and auto-tune is just one of them and it's the one that's really popular at the moment. If you feel that it sounds unnatural, I don't like unnatural, that's fine. But it just is another way of singing. And I think that this criticism is fair. - It's a bit overblown. - Okay, I guess I agree to a large extent, but I do think there is something that I just can't sort out inside of me that there's something about autotune that feels like hiding. That it's like, if you're, if you're, it's, 'cause I actually, I buy the thesis that there are things that sound like autotune, but I do think, and maybe I'm wrong about this, that there is still a real use of autotune that is not designed to be seen, right? Like, if you're right that 90 plus percentage of music is using autotune in some way, like, I don't know. Taylor Swift is not somebody people like, she sounds like autotune, but she's using it, and there are people who have really pissed about that, and because they're like, oh, it's dishonest in some way, that this isn't actually what you sound like. This is you, this is the equivalent of like putting a face filter on yourself before you post a picture, which is now another thing people have sort of visceral reactions to online in this moment of like fake authenticity. And I don't know, like, do we need, is there like, is there like an autotune watermark that should exist on every song that uses it? I don't know. - Okay, so you're getting at this question of like, authenticity just with the pop star, like we want, something about us wants our pop star to be sitting right next to us, our best friend, telling their personal stories, and we know that they're recording it in a professional studio, working with producers and songwriters to make this presentation, which is exciting, and it enhances all of our emotions. So there's always this conflict about what is real and unreal and recorded music. And I think especially in pop music than other forms, we don't expect our actors to be themselves in a film. And yet pop stars are putting on a performance, and yet we have this expectation of they are, who they say they are. You brought up Taylor Swift, for example, and Taylor Swift uses all kinds of vocal processing in her music. On the song "Delicate," she uses a vocoder. On the song "Bidnight Rain," she uses a tool called "Formant Shifting," where she sort of shifts the voice to sound more masculine. I don't like the moments in Taylor Swift vocal, where I can hear audible pitch correction done poorly. And I've heard it done one or two times. Sometimes when you're producing a song, and you need to get it out fast for whatever reason, getting these pitch tuning tools to sound natural actually is a skill. You can get paid a good amount of money just to do the really slow version of hand-drawn auto tune. This is a big part of the craft. And what you're trying to do is make something which was a beautiful, emotional performance. Be a little bit more in tune. And if you do it wrong, you hear this weird artifact. It's unpleasant. I don't like when I hear that, because it's the attempt of a naturalistic performance, clearly being manipulated. You have to hide the tools you're using to clean things up in order for it to sound good. In the same way that a Photoshop photo that is poorly done and you can see some kind of weird, like the anatomy of the person isn't right, we don't like that, even though we completely accept that every single magazine photo is photoshopped. - Interesting. So there's a real, as long as you don't make me think about it, I can rock with it, kind of thing. So it's like as soon as you put this cognitive dissonance in front of my face, it's gonna feel bad. But otherwise, I can sort of internalize that it's happening. I just wanna have to think about it. - It's the same thing with CGI, right? So I just watched Alien Romulus recently. And this CGI of the Android character is trying to recreate a deceased actor and it doesn't work. It's so obviously inhuman. And even though they're in Android, it just like, it doesn't look right. But the rest of Alien Romulus is like full of CGI. And at no point, I'm like, not Alien, just doesn't like feel alien enough to me. So it's all about the context in this presentation. Anything which is supposed to be completely authentic entirely natural that is processed, we will perceive as that doesn't feel right. Things that are presented as I am a pop star on a stage and everything about me is like perfectly composed. We accept that there is, we know, accept. We expect that things are going to have been enhanced. So I just wanna give everyone permission to be mad at auto-tune at certain moments. But when it's intentional, like go of it, whatever. If someone's trying to do it, let them do it. If someone's doing it poorly, fine, criticize them. - Yeah. - I like it. - All right, we gotta take one more break and then we're gonna come back and talk about where we go from here. We'll be right back. (upbeat music) - This is advertiser content from Comcast Business. As cyber threats become more and more machine-based and driven by AI agents, those of us in the cyber defense space were having to invest at the same pace. Hi, I'm Chris McFarlane, Chief Development Officer at Comcast Business. So when we think about the threat of cyber security attacks to businesses, they're just a lot more automated. They have the ability to think, they can scale literally anywhere from 100 to thousands of time faster than a human can. The reality is that today, you need to protect every device, every endpoint, every workload that's running on a server, whether it's your own server in the cloud. This is an area that Comcast Business excels at. We're enabling intelligence-driven defense with integrated visibility, AI-powered threat detection, and automated response capabilities that prioritize, correlate, and contain risks before they escalate. And so it doesn't matter whether you're a small business or a medium-sized business or a really large enterprise. We've got this incredible portfolio of cyber security solutions and services to really enhance your cyber security posture. To learn more about how Comcast Business can help protect your business today, visit ComcastBusiness.com/Cyberscurity. (bell ringing) All right, we're back. Hello, Charlie. We've gone long enough without talking about AI videos on YouTube. We've already had videos on YouTube. So I want to, I actually want to talk about basically two different pieces of where I think things might be headed from here. One is AI and one is back to the kind of authenticity thing. Let's talk about AI first. And I bring this up because I think there is a possible future for AI that looks a lot like the arc of auto-tune that you just described, where tool becomes available. Some people use it like a tool and it's rendered mostly invisible and it helps in the process and it becomes a tool in which we'll get. Other people will use it incredibly aggressively and showy and it will allow new people to do new kinds of things, ultimately some people will be pissed off about it, but eventually it will just be rolled into how the music industry works and may even become like the dominant aesthetic in the way that auto-tune is. Could be totally wrong, but there's something about that rhymes to me with where AI might be headed. What do you make of that comparison? - Well, we'd have to break AI down into all the various ways that it's being used for music because AI is just sort of like a process that can be used in lots of different ways from writing lyrics to separating stem recordings so you can separate the bass and drums and vocals and instruments from a recording to generating whole songs for you from a prompt. There's so many different ways that AI is being used. Some of which are maybe more invisible. For example, like if you use chat GPT to help you find a rhyme, there's no way that I can identify that you did that. Except for this, sometimes chat GPT makes really bad rhymes. It's getting better and better. It really likes to rhyme like do and do. And it's like great job, chat GPT. - Yeah, that's the thing. AI songs and poorly written songs use too many perfect rhymes. They're kind of just an overly sweet, really bad single syllable perfect rhymes. That's the tell. But I don't think we've seen a tool yet that is marketed as an AI tool that has become a predominant sound in popular music. There's all kinds of AI tools that I use in my music production, but I don't think of them as AI. So like I use a diver plug-in, if there's too much reverberation in a recording, and like technically it's using a neural net, but it doesn't really have a sonic fingerprint in the way that auto tune does. So I imagine there will be some kind of effect that will use some kind of AI neural net in its thing that will become a sound. But the AI as a methodology is sort of too broad to say that, well, we're gonna hear AI in the future. Does that make sense? It does, and I think I hadn't really thought about it like that until you put it like that, but it is, maybe the strength and weakness of AI is that it doesn't have a mark in its way. Like you can make an AI song that sounds like anything, but it's never really gonna sound like itself. In a way that is, I think by design, right? Like all these tools are meant not to sound like themselves. They're all trying to mimic something else, and they're all generally doing a bad job of it. And that maybe it would be more interesting if they had, like if you and it made a song with Suno, that it sounded like something, and that maybe there is something to that that is actually missing from these tools. - So you know how like if you try to currently make an image of a person in Dolly 3 or mid-journey, like it kinda has, there's a style. There's a default style, and obviously there's all kinds of way you can get around that style, but it's this weird, somewhat cartoony magazine glossy, very thin, like very white. - Well, and there are the things, right? Like you look at the fingers and you look at the lapels, and like they have their tails in that kind of way. - They have their tails exactly. So right now, if you use tools like Suno or Udo, they do have their tails. And perhaps, you know, in 20 years in the future, people like I wanna use classic Suno, because I'm a little. like the things that it does wrong. - Wait, what are those tales? Like, how would you describe those? - Okay, so first of all, a lot of it's trained on actually really low sample rate and bit rate MP3s, just to like reduce the size of the data. And so there is this like grainy, hissy sound. Like it sounds like a bad MP3 from 1997 being played in Win-Ap. Or maybe more relatable for folks would be like, if you turn on AM radio, AM radio has a sound or it sounds like not quite high fidelity, there's a low fidelity sound. The sounds of instruments are strange. Like, okay, take for example a horn section. If you try to make a funk song with horns, the great thing about a horn section is nobody hits the note at exactly the same time, at exactly the right pitch. The excitement is the subtle differences of eight horns, all trying to do their thing at almost the exact same time. And that actually, that inhuman imperfection is what makes it sound really good. AI oftentimes just like makes it all too perfect. Same with vocals. The vocals sound actually pre-tuned. And so you can identify these things. In addition to the vocals, the lyrics are, have way too many perfect rhymes, et cetera, et cetera. I actually assigned my students to NYU. One of their assignments was go to Suno, prompt it to write a song. Grade that song based off of everything that you've learned in class. Now write your own song from the same prompt. And all of my freshman students wrote significantly better songs than the Udo and Suno song. And most of them threw them out entirely. I asked them to try to incorporate elements of them into their track. And they're like, it's too hard, it's too bad. So, but maybe this badness could be something that we like in the future. - Yeah, and that comes back to the other thing I want to talk about with this authenticity thing. I think a theory that I'm starting to hear more and more from people is that the next phase of culture in so many ways is going to be a spin all the way back to, again, whatever you want to think of as, like quote unquote, authentic stuff. It's called punk rock. - Yeah, it's right, and people are going to deliberately start releasing vocals that sound worse because they are more honest somehow. And we're going to go back to, you know, not using the internet in the ways that we've been using it. Like this idea of just sort of excuse all of this stuff because it has made us not ourselves anymore and going back to like, I don't know, I'm going to just put a bunch of instruments in a room and record like the Beatles 60 years ago and that that is the next thing. - Who famously used the music studio for all of its wild creative tools and use all kinds of post-processing vocal tools, like Varys Beat and et cetera. So I'm just saying who've always been doing this. But yes, I hear where you're going. - Again, again, the question of like how far back do you have to go to find something quote unquote authentic is a good one. - We're all going to play the mandolin, guitar and banjo and violin and, you know, sing and old-timey bands is what you're saying. - Yeah, and like if I sang flat on the take, it's going to be flat on the record. And I would say historically speaking, there have been a lot of people who want us to go that way and we have never ever gone that way. But I do think there's a reasonable argument that AI is like the end of a certain road of inauthenticity that maybe we are due for pushback the other way. - I think that that's a completely acceptable expectation. Like we've seen that trend many times in the history of popular music, in the rise of electronic recording and the electric guitar, louder sounds and rock and roll, you had the folk resurgence. After the super highly produced era of the 1980s, you had grunge as a pushback to that. So I think it's completely likely that we're going to see new sounds and styles emerging that do try to harken to our most human sound as a pushback to anything that is AI. At the same time, I'm wonder if we've already reached peak authentic because if you go on TikTok today, I feel like every other thing on TikTok is the TikTok shop, is some sort of advertisement and it's all filmed the exact same way. It doesn't matter if it's a creator with 10 likes or if it's progressive insurance. It's all like, hey, I'm holding my phone, I'm backlit. I'm just like talking naturally, just shooting the breeze, whatever. Oh, and by the way, this is progressive insurance. And so it's already been co, that like the authentic human thing in video has already been co-opted by advertisers. So where people go next, I think it might be weird. It might be something like auto-tune that is the effect of the next generation of music. - Do you have any theory about what that might be? In this world, what are the teens doing that is gonna seem crazy to the old folks in a couple of years? - They're still using a lot of auto-tune. It's not going anywhere right now. It seems like we have a joke about everybody playing the mandolin and there does seem to be a luminous style of resurgence and the music of "No Acon" is a great example. ♪ The firm on the pitch deceives another sticks ♪ ♪ And I chalk your mom, she forgot that I exist ♪ - I was gonna say, thank you stick season. - Yeah, exactly. So I don't think there is the emergent thing yet. People are still having a lot of fun making music on their computers because it's the easiest way to record. So I think we're seeing more a like, branching of many different kinds of expression than a solidification around one sound. - Can I tell you my theory? - And I'm very curious to know what you think of this. My theory is that the voice memo demo is going to become an actual honest to God, like genre of music. - Oh, interesting. - I think that you see this all the time, right? Where like a song will come out and then like, Charlie Puth does this all the time. He'll go on TikTok or whatever at some late night show and play his demo of the song where he's just like, sort of inventing the melody in his head and trying to get it down into his phone. And that part of me is starting to wonder, like maybe we're going to get to the point where the polished thing and the unpollished thing actually get to live next to each other. If instead of there just being sort of one canonical finished version of the thing, what we actually want is lots of different experiences of this same kind of thing. All the way down to just like the artist riffing into their phone for a minute and that like we want to experience all of that as fans. The new Halsey record has multiple tracks that are exactly that, that are basically like voice memos and demos and things that are interspersed. And that has a long history of like interlude tracks on hip-hop tracks or kind of like, oh, this is a phone call. But I like that. I mean, I want to just point out one other trend that has popped up as well as like, the return to 1920s, 30s jazz or like, what you might call classical pop with the artist Louis Vey who are making things that are a false nostalgia for a music which is 100 years old and that young people are into. So yeah, there's all kinds of routes. I don't know if I really want to hear a lot of people's voice memos that I've heard. - As we look at, you know, we're 27 years into auto tune, which is a long time in music genre and pop music. Do you think if you fast forward a while, we will look back at this as kind of an auto tune era in the way that we've had sort of distinct errors in music. And it'll linger, but like the idea of auto tune being kind of the dominant sound of music will go away and will be replaced by something else. And this will be the auto tune moment or is this just going to be how music is forever? - What's amazing about all listeners, whether they're a trained musician or not, is they can turn on a recording and often tell you, oh, that's like an 80s thing and I don't like the 80s. And the reason why they can do that is because snare drums were produced in a very specific way for the 1980s. There's lots of other production techniques that sound 1980s. And every era has these sonic artifacts of their moment and you can, and they place them in time. The same way that a point and shoot Kodak photograph from the early 2000s, screams early 2000s. You can even have AIs generate things that look like point and shoot Kodak photographs, which are actually having a comeback. I've been at birthday parties recently, where people are handing these things around to get that aesthetic. - Yeah, people are out here buying cool picks cameras from 2003 again. - Yeah, exactly. And so I think that 20 years in the future, it may be that auto-tune is no longer the predominant sound. I think that's probably likely. But when someone's like, but I want to sound like 2008, they're going to use that again as an act of sort of creative nostalgia. So I think that we have been in the auto-tune decades, we'll probably move beyond it. But in the same way that like the electric guitars, not the most popular instrument today, but it's still ubiquitous in recording, I think that auto-tune will be around in some kind of form and it might just sort of fade into the background and pop forward later on, who knows? - Okay, but right now we're still very much in the auto-tune era. - I mean, I think the coolest record of the year was Brat by Charlie XCX and she admits that she has no longer has the ability to sing without auto-tune because it's just so part of her sound. She's like, I don't sound good with auto-tune anymore. And that record is amazing, yeah. - Yeah, if auto-tune is what it took for us to get Brat, I'll take it, we'll be all right. - Oh man. (laughing) - All right, Charlie, thank you so much as always. This was really fun. - Thanks. - All right, that's it for the Vertcast today. Thank you to Charlie again for being here and thank you as always for listening. There's more on everything we talked about. In the show notes, I highly recommend the pitchfork story on the history of auto-tune. There's a ton of good stuff in there. I'll also link to Switched On Pop. So many. great episodes of that show that you should listen to truly cannot recommend it highly enough. As always, if you have thoughts, questions, feelings, or other auto-tune songs that belong in the canon of auto-sune over the years, please tell me. You can always email us at [email protected] or call the hotline 866-verge-11. We love hearing from you. We have a bunch of fun hotline stuff coming up between now and the end of the year, so get your questions in. This show is produced by Liam James, Will Poor, and Eric Gomez. The vurchcast is a vurch production and part of the Vox Media podcast network. We'll be back with your regularly scheduled programming on Tuesday and Friday. Lots of news, lots of stuff going on, lots of fun things to talk about. We'll see you then. Rock 'n Roll. Off the top of your head, give me a intro to auto-tune playlist. For somebody who is like, "I want to experience auto-tune." What are some songs you would just recommend off the top of your head? Okay. I would first go to other vocal processing techniques, and you'd have to listen to like Crophworks trans-Europexpress using the vocoder. You have to put harder, better, faster, stronger from Daft Punk, not auto-tune technically. You have to listen to some talk-box like "Show Me The Way" by Peter Frampton, "Living on a Prayer" by Bon Jovi, "California Love" by Tupac and Dre. Then obviously you need to have lollipop on there because what is auto-tune without lollipop, you need what else has to go on there. Certainly Travis Scott Highest in the room, the entire album of Brat, but if anyone's song, I'm more of a 365 than a 360, so I'm going to say 365, Drake's in my feelings, Sharers Believe, T-Pains I'm Sprung, and I'm not going to recommend Maroon Fives, she will be left. You also left out blue by Eiffel 65, which I found personally outrageous. No, that's because, I mean, that is, I would have to have like a consumer warning on that playlist if you put blue on there because it is a devilish earworm, but I did say Lil Wayne's lollipop, so you know, proceed with caution. How many third party vendors does your company use? 20, 200, thanks to AI, someone on your team probably added three more this week, and your security lead has no idea. Traditional third party risk management can't keep up. Vanta gives you continuous coverage across every vendor, automatically, so you actually know what's in your stack, and what to do about it, AI on, risk off, vanta.com/tprm.