AI – The Assistant Not the Replacer

AI is reshaping industries, but how is it affecting audio outside of Voice Replication? This week, we’re diving deep into this “can of worms” on the latest episode of the Pro Audio Suite podcast! Join us as we host a riveting discussion with MPA from WAVES as we unpack:

  • AI’s role in enhancing creativity rather than replacing it. 
  • WAVES AI investment strategies. • The potential risks of letting AI scrape and replicate without human creativity.
  • The balance between traditional mixing and AI’s edges in audio engineering.
  • Perspectives on the rise of AI voices in media.
  •  
  • We’re not looking for a future where AI spits out a Taylor Swift song on command. Instead, we’re all about how AI can serve as the ultimate assistant in the studio, ensuring creativity and passion remain at the heart of production. Don’t miss this nuanced conversation on the future of AI in pro audio. 🎵 Be sure to subscribe and join our Facebook community for more insights!

    #VoiceOverTechTalk #ProAudioSuite #DesignSimplicityInAudio

    A big shout out to our sponsors, Austrian Audio and Tri Booth. Both these companies are providers of QUALITY Audio Gear (we wouldn’t partner with them unless they were), so please, if you’re in the market for some new kit, do us a solid and check out their products, and be sure to tell em “Robbo, George, Robert, and AP sent you”… As a part of their generous support of our show, Tri Booth is offering $200 off a brand-new booth when you use the code TRIPAP200. So get onto their website now and secure your new booth…

    Tri-Booth

    https://tribooth.com/

    And if you’re in the market for a new Mic or killer pair of headphones, check out Austrian Audio. They’ve got a great range of top-shelf gear..

    https://austrian.audio/

    We have launched a Patreon page in the hopes of being able to pay someone to help us get the show to more people and in turn help them with the same info we’re sharing with you. If you aren’t familiar with Patreon, it’s an easy way for those interested in our show to get exclusive content and updates before anyone else, along with a whole bunch of other “perks” just by contributing as little as $1 per month. Find out more here..
     
     
     
    George has created a page strictly for Pro Audio Suite listeners, so check it out for the latest discounts and offers for TPAS listeners.

    https://georgethe.tech/tpas

    If you haven’t filled out our survey on what you’d like to hear on the show, you can do it here:

    https://www.surveymonkey.com/r/ZWT5BTD

    Join our Facebook page here:

    https://www.facebook.com/proaudiopodcast

    And the FB Group here:

    https://www.facebook.com/groups/357898255543203

    For everything else (including joining our mailing list for exclusive previews and other goodies), check out our website

    https://www.theproaudiosuite.com/

    “When the going gets weird, the weird turn professional.”

    Hunter S Thompson

    Summary
    In this episode, “The AI Conundrum in Pro Audio,” the Pro Audio Suite team, featuring George Wittam, Robert Marshall, Darren ‘Robbo’ Robertson, and Andrew Peters, grapples with a hot topic shaking the foundations of the audio industry: the impact of Artificial Intelligence on audio plugins. The conversation spirals into a complex exploration, examining the role AI should play as an assistant in the creative process, rather than a replacement. Provocative insights are offered about the future and ethics of AI—such as the danger of AI scraping from AI without recognizing human input, creating potential issues like “patent trolls” in music. The team also touches on the broader implications of AI in society, including its recent prevalence in content creation, from YouTube narrations to social media interactions, questioning the authenticity and credibility of AI-driven content. This episode delves into the massive “can of worms” AI opens, from personal AI chatbots to AI voices possibly infringing on personal identity, spurring a rich discussion on how to navigate the balance between innovation and authenticity in the ever-evolving world of professional audio. Join the thought leaders at Pro Audio Suite to unpack the complexities of AI in the audio realm.

    #AIinAudioProduction #PluginInnovations #ProAudioSuitePodcast

    Timestamps
    (00:00:00) AI’s Impact on Plugins: Introduction

    (00:00:12) Welcome and About Us

    (00:00:42) Gomez on AI and Plugins

    (00:13:10) Michael Questions AI’s Role

    (00:19:57) Artists’ Use of AI in Mixing

    (00:26:04) Public Awareness of AI Tools

    (00:31:26) AI Voices on Social Media

    (00:33:53) Personal AI Projects

    (00:38:39) AI Narration and Voice Borrowing

    (00:44:57) Gomez’s Favorite T-shirt Color

    Transcript
    Speaker A: We’re all talking about AI, but the question is, how will AI affect plugins?

    Speaker B: Can of worms.

    Speaker A: A can of worms.

    Speaker C: A big can of worms. 54324. Let’s go.

    : Welcome.

    Speaker C: Hi. Hi. Hello, everyone to the pro audio suite.

    : These guys are professional.

    : They’re motivated with tech. To the Vo stars, George Wittam, founder of source elements Robert Marshall, international audio engineer Darren Robbo Robertson and global voice Andrew Peters. Thanks to Triboo, austrian audio making passion heard source elements George the tech Wittam and Robbo and AP’s international demo. To find out more about us, check the proaudiosuite.com learn up, learner.

    Speaker C: Here we go.

    Speaker A: So, MPa Gomez.

    Speaker C: Actually, gross.

    Speaker A: It is based on our prior conversation before we got on air, probably the wriggling out of worms and stuff. It’s probably not ideal, really, but yeah. So, Gomez, what do you think? Or how do you see AI and plugins moving forward in the next couple.

    Speaker C: Of years without giving anything away? Because obviously I have an inside of you, one of the largest, well, the largest plugin company in the world and the one who created third party plugins because we were the first we have in the past and continue to invest an absolute matzah in AI because the fact it’s just not going away. And it’s very, very exciting. Some of the things that can be done with AI, one of the things that’s really important to us going forward is that we don’t take the approach of AI. We’re going to do everything for you. You don’t have to lift a finger. That’s not a direction, as far as we’re concerned, that promotes passion in music, passion in what you do and passion and effort and creativity when it comes to how you’re building your compositions or your music productions or your soundtracks or your game audio or whatever it is you’re working on. There’s a reason that we’re all in this industry, and 99.9% of it is because we enjoy it. So creating tool that somebody just broke their microphone.

    Speaker A: That was Robert.

    Speaker B: I think I got a dodgy mic. Cable. Sorry.

    Speaker C: Yeah. One of the things that we want to make sure of is that we focus on how can AI be the best assistant for what everybody is doing, not replace what they’re doing. It’s no fun if you type in, write a country music song in c minor, make it about a guy losing his new electric rivian truck and then a girl coming into his life and making him happy, and suddenly it pumping out a Taylor Swift song, there’s no fun in that there’s no journey in that there’s no ownership in that the experience is gone. So what we’re trying to do with AI, and I believe most companies are doing the same thing, is learn and work out all of the ways that AI connected to plugins can be more like jeeves or whatever it is to the robocop, to Iron man, where it’s like the assistant that knows everything and is always there, but doesn’t necessarily do everything for you. And this comes back to something that’s really central with anything to do with artificial intelligence. Whether you’re doing something with ChatGBT or Claude from, you know, Gemini Pro, or one of the llamas from Meta or Google, it’s already been very confirmed over the last year and a half that if you don’t know what you’re doing or you don’t know what you’re talking about, then you can’t get the most out of AI. Because AI does have a tendency to work better for people who actually can see and understand and perceive what is correct and what is not correct. When it comes to the outputs, some of you may have heard or may not have heard of this. If you’re listening to this AI hallucinations, the AI hallucinations that happened with Chachibt initially, and then with Bard from Google were epically amusing to those of us who were beta testing, but terrifyingly convincing. When you have an AI, write an essay on something that sounds so much like it actually happened, but then when you fact check it, you find the names don’t exist, the history dates don’t exist, none of it exists, but it gave you roughly what you asked for. But if you don’t know the topic, then you don’t know that it’s wrong.

    Speaker A: Correct.

    Speaker C: And one of the things that we’re on that line of right now when it comes to AI, creativity and plugins, is how do we continue to evolve this technology and large language, large learning models and tensorflows, and all these different wonderful things, so that they assist without creating stuff that’s completely in the other direction, and we have to dig ourselves out of a hole. And I think what’s going to happen is you’re going to end up with plugins that are more like mixing assistance or composition assistance, or sample creation assistance. In fact, there is already one, there’s one called, I think it’s called synth GBT that you type in. I’m looking for a bell sound, and it will give you a bell sound, like chibla belt or it’s still fairly rudimentary, but if you look at how fast things move, it’s not going to be long before you can put in. Okay, so I’m after a 16 pomophonic pad that evolves from this cut off.

    : Starts in a minor chord, moves into a major chord over the course of blah, blah, blah, blah.

    Speaker C: Exactly. I want it in this BPM, and I want it to duck like a pumper, like it’s side chained to a kick drum, and have it pump something out. Now, the other problem with all of this that’s going to come up the same as it already has done with image generation and text generation, is copyright. Because at this point in time, it’s been determined that if something is created by AI, then human being does not have copyright to that. In the music world, that’s going to create all kinds of fun and frivolity.

    Speaker B: Can you imagine?

    Speaker C: Yes, I don’t need to imagine. It’s already happening. One of the things that we’ve already done with things like studio verse, which is part of our rack plugin, that studio rack is you can scan your audio, and we will automatically, using AI, suggest a preset chainer for that track of audio that you’re putting in. Whether it’s a bass guitar or a vocal or a female vocal, will suggest some options for presets. The other thing you can do is you can type in into a new thing that we’ve got coming out, which is like studio burst for vis. You can create a chain of vis, or type in what you’re looking for. And we’ll suggest a Vi chain with macros, with automation, with levels, et cetera. That is pretty much created by a mixture of existing technology and AI. And we’re finding that it’s getting easier. And if we’re finding that, then other plugin companies are finding this too. It’s getting easier and a lot faster to create these things. The challenge is the easier it gets, the harder it is to create something that’s actually going to be a wow factor for somebody that’s sitting in their room and working on their music until 04:00 in the morning. You can’t just give them AI anymore in anything. It’s got to be epically good at helping somebody be a better version of themselves.

    Speaker A: Interestingly, I look at this, and you go through all the big evolutions, and we’ll use music as an example. But if you think about the things that changed everything, first of all, electricity changed everything, a microphone changed everything, a speaker changed everything. So you got an amplifier, all these different things that everyone’s sort of gone. Oh yeah, electricity, electric guitars, microphones, amplifiers, all that kind of stuff. In the old days, if you would stack your instruments based on the volume and have one microphone in a room, then all of a sudden you’re micing cabs in a studio and stuff like that. You got multi track instead of one sort of monotrack. All these things have changed. But this is the one that’s kind of a bit scary because this is.

    Speaker B: The one that takes the personality out of it. I think that’s the thing.

    Speaker C: That’s the risk.

    Speaker B: The risk is there that you take the humanity out of it and it just becomes this robotic thing.

    Speaker C: There are so many websites already that will let you type in a topic or type in some words and it will pump out a rudimentary rap or pop song for you in some kpop style. And they’re all awful.

    Speaker B: Somewhat argue they all have anyway. Gomez right, but they’re all awful.

    Speaker C: But one of the things that needs to be said is these websites and these AI models that are being used are sitting on a server on a huggingface co somewhere that somebody is renting. In some cases it’s a 17 or 15 year old boy or girl writing this code in an evening and then asking AI to help it with the front end interface for it and then punching out a website. The fact that it sounds awful is not what we should be paying attention to. The fact that it’s so easy to create, that’s the part that we need to be alert to, because we need to remind ourselves that this is a very slippery road and we could ruin an entire reason for creating if we let AI take over the job that we love doing. I think, look, I feel like I just did an Obama speech or something.

    Speaker B: I’m a bit more on the fence, I guess, because I look at it this way, I think we all needed to start somewhere, right? And if you go back to ask Gomez, we walked into cart prep, me at two SM and you in the days at triple M, and you learned your craft by sitting there and then walking into the studio and eating your ham and egg sandwich while watching Jeff Thomas crank out an amazing promo, right?

    Speaker C: Four BK for me.

    Speaker B: Well, there you go. That’s how we learned our craft. These days, there’s less of an opportunity for that. So for me, I can see that AI provides this opportunity for someone who’s trying to figure out their craft in terms of, okay, well, these are the basics. These are the rudimentary things you need to understand AI gives someone the opportunity to go and learn that stuff. But I think the trick is that we don’t need to. But if you’re going to hone your craft, you then need to move on from that. So I can see that AI provides a starting point, if you like, and it maybe makes it easier for someone to understand, whether it be post production or music or whatever you want to talk about, if it gives them a place to start, I think that’s a good thing. But I think I do also agree with you that it opens a can of worms, of taking the humanity out of the craft as well. That’s fine.

    Speaker A: That’s the whole thing. Yeah, the whole thing really is. That made music really interesting was human frailty. Like, the amount of records that probably your favorite record, there’s probably mistakes all the way through it, and that’s what made it appealing. There was something about that that worked, and when you take that out and it’s become sterile, it just doesn’t feel human anymore.

    Speaker C: If you take people like Han Zimmer or to another extreme, Moby or lemon stellar or Frufru or, they all created their own style, and that style wouldn’t have happened if it wasn’t for them creating what was in their heads at the time. And this is one of the things that, I mean, when I say that if we let AI replace everything, then what will happen is what’s already started to happen, to a point where we’re finding that AI is skewing the information on the Internet, because AI is starting to already scrape information from AI and not know the difference between human created and AI created. So what we already have is a scenario where, I mean, I currently have access right now to 36 different large llms and tensorflows from all the major companies and some of the smaller ones. So I’ve got 36 different ais that do things. And one of the things that we’ve been finding is some of them are already not knowing what’s human created and what’s AI created. And that’s a huge issue, because what it’s going to do, especially in things like the audio world and the music world, is it’s not going to know when it’s ripping somebody off and when it’s taking a style that’s been scraped legitimately or illegally, and therefore it’s going to end up. I mean, things like the court case that Ed Sheeran just went through, which had nothing to do with AI, and had everything to do with a family of somebody saying, oh, you totally ripped this off, whereas he didn’t. Those kind of court cases are just going to get more and more prevalent with AI around because of the same reason that visual artists are already suing the major AI companies. It’s getting super hard to work out what’s the style of somebody and whether that style is worth something in the value, because they created it. If somebody else goes off and creates.

    Speaker A: Something different, maybe that’s going to be the new income stream for songwriters and creative people. Litigation and winning a court case, that’s how you make your money.

    Speaker B: Living off their old.

    : Some people are going to become the patent trolls of music.

    Speaker C: Elon Musk’s wife has already done this by the. She’s. I can’t remember her name. It’s something. She’s. She’s created an AI virtual model of herself, and anybody can create music with her voice, but she earns a percentage of the royalties off that.

    Speaker B: Right.

    Speaker A: Do you know, that’s funny that this is going way off track, but I saw an interview with Dolly Parton and.

    Speaker C: Oh, yeah, no talent to talent. You’re absolutely right. It is way off track. Yeah.

    Speaker A: But it’s interesting in the sense what you just said about Elon Musk’s life is that she got a call from Colonel Tom Parker saying that Elvis wanted to record, I think it was, I will always love you, or something.

    Speaker B: It was, I will always love you.

    Speaker A: And he said, but we take a cut, we take 50% of the royalties. And she said, I’m not giving the royalties away to anybody, and said, no. And then I saw her say, I’m glad I didn’t get Elvis to do it, because Whitney Houston did a much.

    Speaker B: Better job, and I made money on that.

    Speaker C: Yeah.

    : Well, I think I have a different direction with AI or just music composition in general or style in general. You were talking earlier, Michael, about how these different artists that you named carved out a style, and I think about how those styles are partly shaped by the tools that are available to those artists at that time.

    Speaker C: Absolutely. And the people who have come before them. Yeah.

    : You look at, I think, of craft work.

    Speaker B: Right.

    Speaker A: I was about to talk about work. Exactly.

    : For that reason, their sound is incredibly sculpted on the technology that was available at that time and their unique style.

    Speaker B: Yeah.

    : And art of noise. My God. Yeah. So now you’ve got Hans Zimmer, who arguably has a limitless tool set. Right. I mean, it is. Let’s face it, Hans Zimmer’s tool set is limitless. You can get anything, afford anything. His only limit, literally is know, if it wasn’t for the fact that a film score had a distinct know, that’s his only limit. Right? So now the ability to sculpt something that would have been impossible to do by Hans Zimmer within a deadline, he’s going to be able to expand that by factor know an order of magnitude by using AI Han Zimmers to help him expand the.

    Speaker B: Just.

    : I can’t even fathom where that goes. Know? But it is interesting.

    Speaker C: You know what I’d really like to do? I had this conversation with somebody at, like, 03:00 yesterday morning. I would love it if we could just give. If I was going to give AI all the potential of AI in the music world to one person and say, mess with this and let us know what we can do with it, I would give it to collier. Yeah. Yeah. That kid is nobody else. Just Jacob. Jacob, go off, do your eccentric stuff, work out what we can do with this as humanity, and then come back and we’ll decide whether it’s actually worth it or not. Because that mean. I don’t know how. I didn’t know about Chek Collier until about six months ago, but I am obsessed with how much talent, raw talent that guy has.

    : Yeah, it’s remarkable. I’ve seen him live three times. I saw him at the.

    Speaker C: It’s like. But it’s like I watched a masterclass with him about how there is no such thing as a wrong note. And it’s all about the tonage of how you play the notes. And I’ve never looked at piano the same since. It’s like just this ball of energy that just makes such magic.

    : Absolutely.

    Speaker C: Yeah.

    Speaker B: Here’s one thing that interests me, Gomez, and you might or might not be able to answer this, but you work with these guys fairly regularly at the top. So let’s talk about Chris Lord Algae and that ilk. Algae. Sorry, excuse me. That’s my Aussie Twang. How much of this stuff are they. When I say stuff, I mean AI. How much of this stuff are they using in their mixes? Are they sticking to the old school, or are they sort of going. I can see a few advantages to moving towards this AI sort of thing. So some of it’s sneaking in, or are they steadfastly refusing to go that direction?

    : Yeah, I mean, some artists are way more into trying the bleeding edge, and others are much more into sticking.

    Speaker B: I’m kind of thinking of the Beatles singing into teapots and all that sort of stuff. I mean, for me, it’s sort of a progression. But I’m interested from Gomez, if he can answer the question, how much?

    Speaker C: So the person I had a conversation with at 04:00 the other morning about Jacob Collier was my man, Eddie Kramer. And Eddie, one of his most famous moments that had nothing to do with Hendrix or Led Zeppelin, recorded the Beatles at Abbey Road because he was the only one available and they wanted to record. And one of the things that Eddie was there for at the beginning was this epic use of, okay, so we have one channel and another channel, and that’s it. We’ve all got, how do we make those work? And he was one of the ones who made it work and turned rock and roll into. He didn’t define rock and roll, but he was there as part of the process and as I like to call it, part of the band, in a way. And he’s one of the ones who is him and I will talk about AI until the cows come home, because it’s exciting for him. He looks at it, he’s just turned 80, and he’s like, the potential on what we can do with music with this kind of technology, as long as we don’t screw it up, it’s beyond epic. It’s exciting. Now, if we get back to the clas of the world, Chris is really good at choosing what he has on at the time. I mean, on one side, he just mixed the new green Day record, and it’s already creating huge waves and hits everywhere it goes. The way he did that was they went back to the roots, and they did it exactly the same way. He recorded the first green day record nearly completely analog SSL console to tape as well as to digital, but with pro tools. It’s like they went, okay, so let’s go back to the roots and let’s do it that way. And he was starting to do that when I was sitting in the studio with him in June of 2022, just before I’d come all the way back to me roots. And he was talking about, it was like, no, we’re going to go back to the way it was. And I said, why? And he goes, because it worked. And we’ve got so into this technology that somewhere we lost the sound of that band and we need to find it again.

    Speaker B: Definitely.

    Speaker C: Yeah.

    Speaker B: Yeah. Imagine recording Led Zeppelin with.

    Speaker C: Huh?

    Speaker B: Imagine recording Led Zeppelin with AI. Exactly.

    Speaker C: Well, the way Eddie and I were talking about it was, I posed the situation to said, what if? What if we could use AI in such a way that, because there are still recordings of Zeppelin and there’s recordings that he’s working on right now for the Hendrix estate of cassettes of Hendrix nobody’s heard before, where there are dropouts. Now, these are not like isotope capable dropouts. These are dropouts for like 20 seconds. But what would it be like if you could say, okay, right, so this is the catalog of this artist. This is the problem we’ve got. We need to fill this 20 seconds with what they would auto fill.

    : I need this piece.

    Speaker C: Yeah.

    : And then say, do four versions.

    Speaker C: It’s literally like the. It’s like Adobe Photoshop. It’s like Dolly. Yeah.

    : Extend this image. I need to extend. It needs to be larger, fill in the background.

    Speaker C: It’s exciting to me. But again, we have to really be careful about number one. Firstly, I can pretty much confirm AI is not going to take over the world. We are not all going to be slaves to it. It’s just not going to happen. And if it does, it’s not going to happen in our lifetimes. But what we do run the risk of is we run the risk with being too blase with this technology, and we need to be very careful and present about how we want to use it. One of the things which I’m doing right now, and it takes a lot of time, I’ve never done it before. I’m writing the AI policies for an entire audio software company. And it begins with the word W. I work for them, but I’m writing the entire policy on usage of AI, whether it’s sales or product specialists or coding or anything else, the entire policy based and down to ethics of AI. And it’s hard because just have Chad GPT do it.

    Speaker B: Nice one, George. Cliche one for George.

    Speaker C: The problem is, when you’re writing an AI policy, you can’t get AI to write the damn policy.

    : No, because there’s no precedents. There aren’t books on how to write a policy around an AI. Yeah, I know.

    Speaker C: You’re absolutely right. But here’s the other thing that I think that we need to be aware of here. 80% of the population, if not more, have no idea that chat GBT or bard or Claude or llama from meta or Po or chat GBT pro or chat GBT teams even exists. They haven’t got to them yet. We’re still in the minuscule percentage of the world who are aware of this. It’s like I use AI every day, but I’m using it to get rid of the menial tasks, and I’m using it to help speed up what I do. So I can do it well. But I have to remind myself how many people in the world, regardless of whether we’re talking audio world, visual, or whatever else, have no concept yet of AI and even software companies. There was a study done inside of Adobe, and think about how many staff Adobe has, and there was a large chunk of people who were like, what’s chatchippy? Wow. It hasn’t got to their desk yet. It hasn’t got to their office yet. It hasn’t been relevant for, and these are people who, they don’t necessarily read tech blogs or LinkedIn. They might go home and read a novel. Or I was about to say, these.

    : Are people that actually go home, and.

    Speaker C: These are people working. And that’s the thing. These are people who, yes, they have passions, yes, they have hobies, but they stop. Their work. Stuff stops outside as soon as they get outside. And they haven’t heard of it yet. I mean, a lot of them have, but there’s a huge amount of people who have never heard of.

    Speaker B: Is that a generational thing, Gomez?

    Speaker A: I would say it’s not generational.

    Speaker C: My mom actually asked me to create a chat bot for her that she could use as her assistant, because I just created a company it’s about to launch, not just in the last six months. I’ve created, it’s called creative ninjas, and it enables you to choose what kind of assistants you want, and then they’ll be in slack WhatsApp teams, et cetera, or on the website. And they are your assistants with the names you chose for them, the personalities you chose from. And she actually went through, and she said, I want you to create one of those for me. So she uses a chat bot AI assistant every day, and she’s 85.

    : Those are the people that really need that stuff. My short term memory has already been atrophied dramatically by the use of technology, not enhanced by it. So it’s only going to get worse.

    Speaker C: George, there’s a plug in for that.

    Speaker B: My short toy memory is being destroyed by 35 years of rock and roll radio. But that’s another story. I think that the other thing that sort of comes to mind to me. And if you want to look for the good side of all this, though, there’s AI involved, but it’s not completely AI. But you look at now and then the Beatles thing that just happened a few months ago. I mean, we wouldn’t have that if it wasn’t for AI. That’s a really sensible use of, like, to resurrect a recording like you were talking about with the Hendrix thing to resurrect a recording like that and then be able to.

    Speaker C: And magical. And I feel so blessed that we got the chance to watch that.

    Speaker B: I know, yeah.

    Speaker A: Interestingly. But then once again, it’s the human aspect of that that makes it work.

    Speaker B: Well, McCartney and Ringo and George, to a certain extent to humanized it didn’t mean we’ll never know what John Lennon originally know. God knowing John Lennon, it could have gone anywhere, but they sort of beatilized it. And because they know they knew John intimately, they were able to do that and turn it into something that could have been.

    Speaker A: Yeah. Interestingly.

    Speaker B: Another number one in 2023. I mean, Jesus, talk about infinite.

    Speaker C: Funny.

    Speaker A: Talking about like Eddie Kramer being able to grab stuff and fill up gaps, if there’s gaps missing, dropouts in tapes and stuff like that. Getting back to the voiceover thing, this is where it could be really interesting if an AI can actually do that. Say for instance, I do a 32nd read. So they’ve got the read, they got the raw read down. Then the client decides, I’m not happy with those couple of words. I want to change, I want to revise the script. Instead of me being called back in to do a reread or a drop in, they would just use the AI to copy my voice and change that.

    Speaker C: Andrew?

    : Yeah, I can do that in descript.

    Speaker C: We can do that with the script. We can do that with eleven labs. We can do that with revoicer. There are multiple models that you can already do that. And I would love it if they were all used for good and not for evil. But I think that this is one of those times as well. You’ve just brought up something that’s already here, which is an issue to me. I hear so many AI voices now on faceless YouTube videos and TikToks and even some commercials on tv and radio. And I think there’s no soul in that and I can tell because I’m human. The problem is a client will pay dramatically less for that. And all they care about is that the message is there. The client of the client, the person who’s listening to the ad. I haven’t done a survey yet, but I think it’d be interesting. How do people in listener land feel about AI voices and are they as convinced of the message or do they hear the message the same? Because I just don’t think it’s going to fly.

    : I think there needs to be some actual market research. I mean, it’s got to be somebody doing it already. But sitting people down, either virtually or in a room and showing them and playing them ads and splitting back and forth between human performers and AI performers and literally just doing the actual work, doing the research and getting people’s responses and how they feel when they hear those things.

    Speaker C: Eleven labs has just brought in something else, which is speech to. Speech to conversion, which means that if somebody decides, okay, the AI voice wasn’t good enough, but I need that voice that I created of Andrew, and I’m going to use the AI version because he’s too expensive and I’ve already got all of this information from him. What I’m going to do is I’m going to read the speech with the right inflections and the right emotion, and then it will turn it into his. That, to me, is something I was like, I just wish they hadn’t got to that that quickly.

    Speaker B: Do we need to go there?

    Speaker C: It’s already there.

    Speaker B: Yes, but why do we need to go there? Because that’s only leading down a dark. That’s. That’s the thing.

    Speaker A: Well, it’s like the thing that Simon Murphy did where he got my voice. He trained his AI machine to copy my voice and then had me singing the lumberjack song from Monty Python.

    Speaker B: Yeah, but see, that was.

    Speaker C: That’s a good use of technology.

    Speaker A: Not if you heard it.

    : Speaking of using technology, I mean, I’m using it. I’ve made my own Chat GPT, custom GPT of me. And it’s freaking awesome. I love it. It’s so great. I can sit there and answer 30 to 40 questions on Facebook and Reddit in an hour and I’m a hero. And they’re great answers. They’re far longer and more verbose and more complete than the answers would be if I had written them. Because I don’t feel like typing all that stuff out and I can’t remember half of it, and it finds all that information and it references where it came from and it gives them what they want and they’re like, thrilled.

    Speaker C: You used GBT Pro for that? Like, just created a GBT.

    : I am using an app that probably. Well, the app I’m using is GPT for turbo or something on the back end, but it’s an app called Custom GPT. Okay. The reason I liked it was because it was an incredibly friendly user interface. I mean, I could understand very quickly. It had very good transparent terms and it had extremely strong language about being anti hallucinating or using anti hallucinating technology, which, if you’re going to make a chat bot, based on language model that you are providing of your own sources. It’s incredibly important that it not hallucinate, right? I mean, I will not dare put something out for the public to access, which this will be for the public to access and subscribe to. And I would never do that if that information wasn’t sourced and sourced of my own content. And that’s what this is and it’s revolutionary. And I can take that same content now in about a week I could have a book made, hardcover, bound book of that information using cheap BT technology to write the book. I know how to do that. How is that helpful? Well, that could leapfrog me into being able to get a speaking gig at universities and that could get me a speaking gig at a TEDx. All that stuff is all part of the journey of getting these kinds of levels of respect and notoriety. And so that’s all on my roadmap now where that was going to be on my roadmap. But maybe in two to five years, now that all that stuff’s on my roadmap in two to four months, right. I mean, it’s just accelerated my goals.

    Speaker B: But to be fair, you’ve got to have that knowledge in the first place. You can’t just do all that.

    Speaker C: That’s the key. You can’t not know what you’re doing before you start.

    : Absolutely.

    Speaker A: Unfortunately, there’s plenty of people out there who don’t know what they’re doing and they still become experts.

    Speaker C: Well, this is nothing to do with AI, but it’s like, let’s not remember anybody out there. If you want to start doing something, start doing it today, because there’s plenty of people who are way more irritating than you ever will be who are already doing it and they don’t have half the experience you do and they’re charging for it, those dicks. So go do what you want to do. Do the things you want to do.

    : No, I wanted to make sure that I had ownership over this, even though I know that I can’t.

    Speaker C: But you don’t.

    : Well, I know that I can’t stop somebody else going out and scraping a lot of that content. Not all of it. Some of it’s totally private, but I can’t totally stop them. But I wanted to be doing it first and have it branded under my.

    Speaker C: Name because that’s not what I meant. What I meant was because of the laws. So far, when it comes to AI, it doesn’t matter whether nobody’s going to stop you putting out a book. Nobody’s going to stop you creating any of that. But if the copyright is ever contested, you will lose, not because a lot of the initial knowledge isn’t yours, but because of how it was put together.

    : No, you have to actually divulge in the book, this was written using AI.

    Speaker B: Oh, do you really?

    : Oh, yeah, right. You can’t publish a book and say it was written by your own intellectual property without AI and falsify that you will lose. Yeah, it’s an interesting, well, talking about.

    Speaker A: That with credits, like in music credits, all those different credits for film, television, whatever. They talk about having a credit for saying blah, blah, blah done by AI.

    Speaker B: Yeah. Well, what if AI narrates a documentary and they borrow the voice from somebody? Do they have to? Is that credit?

    Speaker C: Before I forget. So there’s this guy, I can send you guys the link afterwards. There’s this guy who I just want to sit down and buy a beer for. And he’s a coder, but he wrote a language model and he wrote code so that everything he did on his computer for a full day was narrated in documentary style by an AI voice and character of David Attenborough. And this guy went, they, he went so far with making it so accurate that this David Attenborough, like, he’s sitting on camera now. He looks like he’s confused. He may be trying to work out. It was so exact, it blew my mind, but it was so entertaining. I’ll find the link and send it to you. It happened about a month ago and it just showed what’s possible. It’s like I hadn’t even thought about doing something like that. But if he can do that with his daily life, then you could give an AI voice model like that, a bunch of edited footage of the Amazon or skateboarding or anything else and have it write the script and do the voiceover real time for you.

    Speaker B: But here’s the thing. Gomez, right, let’s say he put that on YouTube and he ended up with 5 million views. Okay? He’s now making money.

    Speaker C: He doesn’t earn anything.

    Speaker B: No, I get that. And I’m not pointing the finger at this guy, but this is a what if, right? So if he put that on YouTube and he gets 5 million views, all of a sudden he’s now making money from David Attenborough.

    Speaker C: Yes.

    Speaker B: So where’s the copyright claim in that? You know what I mean? This is the massive can of worms that gets opened by all this, which.

    Speaker C: Is why YouTube will not let him earn money and not even add money from that video.

    Speaker B: Right. Okay.

    : It’s demonetized.

    Speaker B: Well, it shouldn’t be.

    Speaker C: Yeah, this is one of those things that YouTube has been doing really well and Google have been doing really well for years. And I should know because the amount of events I’ve run for waves, if you so much as a melody or even hit a kick drum that’s in a Dua Lippo song or anything else, you will be flagged and you will lose any monetization on that video whatsoever.

    Speaker B: Think of dua lipa should be taken down anyway.

    Speaker C: With some bands they will actually like Led Zeppelin for example. If you pay more than like think 5 seconds, YouTube are required to mute the video completely. So when it comes to voices, like David Attenborough and the voices. Can you guys hear my dog snoring?

    Speaker A: Yes, I just heard your dog snoring.

    Speaker B: After last night, he’s probably a bit tired, mate.

    Speaker C: What a weekend. Good dog. But yeah, it will recognize the vocal pattern and it will automatically say, copyright flag checked. No monetization on this video. And you can do three of those and you can contest it, but your chance of contesting it is pretty much zero.

    Speaker A: The scary thing is if someone grabs something out of that David Attenborough voice, talking about something and uses it as like, well, this is what David Attenborough said and boom. And then people start believing it.

    Speaker C: Yeah, I can share something which I’m working on, which is not waves working on, it’s me working on it. With all the research I’ve been doing in AI myself. My mother is one of my role models. It’s like what she’s done in her life is epic. She is past middle 80s, she’s never written a book about her life, but I’m starting to get her to put it together. What I’ve also done is, and it was actually one of the reasons why George I bought this mic was so that I could sit across the table from mum and actually get her to talk about everything. While I’ve been doing that, I’ve also created an AI model clone of her voice. And once the book is finished, whether she is alive or not, I’m going to put out the audiobook version of it.

    Speaker A: Cool.

    Speaker C: That’s very cool because I believe that it’s a good use of the technology.

    Speaker A: Yeah.

    Speaker C: My mum has trained equestrian Olympic teams of the US, the UK. She trained the royal family. Princess Anne used to kind of pick me up as a baby and move me back to our tents at all shows when I got lost.

    Speaker B: That’s a princess.

    Speaker C: It’s totally also where I get my hair from 15 all. But it’s like she’s a legend in that industry. She’s never written any of it down, and that’s just a shame to me for the world. So I’m using AI to make it possible for me to be able to help write that book and then get AI to do the audiobook version.

    Speaker B: Yeah, that’s a great idea.

    Speaker A: See, I should have been your mother’s son, because at least now I look like I should be part of the royal family.

    Speaker C: That’s one of the weirdest things I’ve.

    : Ever heard anybody say.

    Speaker C: One of the weirdest things that’s ever happened on this podcast.

    Speaker B: Someone rolled the credits. Before this goes any further, I just want to make.

    Speaker A: All right, we will wind it up. Thank you, Gomez. But I want to just say a couple of things. One is George in the background there, you’ve got an Australia sweatshirt, and I think that’s from Larissa Gallagher.

    Speaker B: Nice.

    Speaker C: Indeed I do.

    Speaker A: And the other thing is, you’re wearing a tributh t shirt with a microphone in the shower.

    Speaker B: Hello.

    Speaker A: To which we’ll never unsee.

    Speaker B: Tributh. Thank you very much for your sponsorship.

    Speaker A: And also, there you go. Look behind me. I don’t know whether you can see it. There’s a tributh there, all illuminated. Do not disturb. It says in there.

    : Sorry, too late.

    Speaker A: Yeah, that’s right. It’s very disturbing.

    Speaker C: No, tributes are great.

    : Yeah, sorry, you’re already disturbed. Sorry, it’s way too late.

    Speaker C: It’s probably one of the only podcasts that’s ever had two tula mics in it at the same time.

    Speaker B: There you go.

    Speaker C: Get in contact with them.

    Speaker A: See if you can get some free shit.

    Speaker C: Is that really green, though? Really greenish.

    Speaker A: Is it stuck at green or turquoise?

    Speaker B: Turquoise or teal?

    : This is sea foam.

    Speaker B: Sea foam.

    Speaker C: I’m sorry.

    Speaker A: I call that color North Sea, I reckon.

    Speaker C: Or the Atlantis, actually.

    : I don’t know what this color is.

    Speaker B: But anything that’s not black isn’t cool.

    : Anyway, I wanted to get something different.

    Speaker C: It’s funny, isn’t it? All four of us are wearing black t shirts. We’re just so old.

    Speaker A: Yeah. No, it’s because it doesn’t look slimmer in black.

    Speaker B: That’s right.

    Speaker A: You can’t see the wrinkles.

    Speaker B: Very rock and roll. You don’t see Angus young wearing sea foam t shirts. Can I just say.

    Speaker C: Well, that was fun. Is it over?

    : The pro audio suite, with thanks to Tribooth and Austrian Audio recorded using source Connect, edited by Andrew Peters and mixed by Voodoo Radio imaging with tech support from George the tech Wittam. Don’t forget to subscribe to the show and join in the conversation on our Facebook group. To leave a comment, suggest a topic, or just say g’day. Drop us a note at our website, theproaudiosuite.com.

     




Check out this episode!

Leave a Reply

Your email address will not be published. Required fields are marked *