Crash Course on Artificial Intelligence in Music w/ AI Music Tech CEO

I grew up a rapper. Making that AI model was the only reason I was ever able to produce my own beat. It really is like a jumping off point. It's like an inspirational point to get started. We work with artists, and you know, we make sure their data set's clean. We pay them for it on a monthly basis, and we don't take away any rights. Imagine you play a melody, and AI immediately plays you counter melodies, B sections, etc. AI, AI, AI. Will it take all our jobs and eventually be the downfall of humanity? I don't know. But in the meantime, there are a lot of cool ways to use it. Our guest today is the CEO and founder of Lemonade, spelled with a sneaky little AI in the middle, which is an incredible tool for producers and songwriters to always be inspired. This episode is all things artificial intelligence and is perfect if you're kind of on the fence about whether you want to start using some of these tools or not. But it's also great if you have already accepted the power of our AI overlords and are just looking to brush up on the state of AI and music from an absolute expert. Now let's get into that sweet, sweet theme song.

One spin up spinning, one set. Have you guys had one by the way? A dog on the podcast? No, no, a samosa. Oh yeah. I was like, a dog on the podcast? Yo, I got that dog in me. Don't worry. Yeah. No, I have. I have had Samosas. They're delicious, actually. Yeah, for sure. Well, MJ, welcome to the podcast. Yes. We see you coming on, man. Oh, these are loud. Okay. Yeah. This is just this is a conversation that we really all need to have. We have to let's get early on this. Like, be early. be early. Like artificial intelligence is just such a buzzword within the music industry and across all industries, to be honest. The world. So I really want to just treat this episode for myself and Henry's sake, as well as for our, you know, community's sake. I want this to be just as much of a crash course on AI in music as possible, right? Because I think you know, it's come a long way, it's gonna continue to go go even further, right?

So I just want to kind of start from your perspective. Like, what was the earliest example of AI in music? Just to like set the stage for everyone, like when did it start? Yeah, for sure. I mean, I think the first thing is before AI and music, it's just like AI is the crazy topic that's going off right now. But it's really generative AI, which is a thing in recent years that people are like, you know, going crazy about of like, oh my gosh, this is either so interesting or it's so scary or whatever. But AI has been around since like the 1900s, actually, maybe even I think the mid-1900s, but either way, so um, and so yeah, generative and and so generative AI for like layman's, you know, people is like a ChatGPT type of solution, right? Yeah, okay. Essentially being able to train on a data set and be able to use that data set to create something completely new, so it's generating something completely new. Okay, cool.

Um, which is different than how AI, like traditional AI works, where like if you're using Siri, for example, and that's why I'm like AI's been around for a while, because like Alexa devices, Google Home, Siri's, list goes on, like Teslas, those are all forms of artificial intelligence, but they weren't necessarily generating things, they were learning human patterns and making predictions, but generating is completely different, right? Like, um, for example, when you use Siri, it's like, hey, who won the game the other day? It's using AI to retrieve information of like okay, the Bills versus the Jets. This was the score. It's not generating a score. But when you use something like Dali, if you guys remember that image generator, yeah, I guess Midjourney's the big popular one now, right? It's been popular for so long. But the overall point is like when you ask it to generate something super specific that's in your mind, it's not retrieving something from like a Google search. It's taking its training catalog and making something completely new, right? So anyway, so so that's the hot topic: generative AI. Yeah, and that's where it gets that's I mean, generative AI is the creepy shit.

I'm gonna be right, like that's the sure that's like generating new things and concepts and ideas and and you know, thoughts, that's a lot scarier for most people than just like it's the AI we've seen in the movies. That you know, is because the villain in the movie is its own thoughts, its own brain. It can kind of come up with stuff, right? Versus, yeah, to your point, like the series, the Alexas, like they kind of just on command, they they got stuff that already existed, right? Exactly. But to answer your question though, so when was the first time like generative AI and music kind of went hand in hand? Um, so now that we know AI has been around for a while, generative AI. So um, do y'all know what the T and ChatGPT stands for? No idea. Treat us like treat us very not non-intelligent about this whole thing. In like an intentional way, because if you guys knew, I would have been hyped. But no, no, no. The majority of people don't know what the T stands for. So the ChatGPT, the T stands for Transformer.

The Transformer was actually invented by Google. Uh, let's see, right over here. There it is. it was invented by Google in 2017. So the concept of generative AI, um, I'm not gonna say it necessarily started in 2017, but the basis of the Transformer started in 2017. Um so the Transformer, of course, was an open source model that anyone could just like use. And um, at least for me, my first time messing around with generative music was using something called Magenta, which is an open source, another open source library from Google, which is based off of that transformer engine that Google created, right? So the whole idea of what Magenta is, um, it's still a team at Google, a research team, but uh it allows you to train your own AI MIDI models and things of that nature. But for the most part, like I would say it was like 2018, 2019 is when I heard my first like inklings of what AI music was. And it was so bad.

Like I think, like the thing, like where it's at right now, it's not great, to be honest, like in terms of fully generative stuff. And I'm glad that it's not great. Uh, because I have a whole like stance around like where AI should come into the creative workflow to try to protect the creators. But anyways, in 2017, 2018 or whatnot, it literally was barely capable of outputting things like twinkle twinkle little star, like that level of coherence. Um, but yeah, so it's been around for a long time. But the thing is, I would argue open AI, when they released ChatGPT, they kind of awakened every AI company out there. You know what I mean? Because if you think about it, Google has had a head start for a long time being an AI company and like having so much in like AI research, and they created the Transformer. But when you use a like large language model, do y'all use BARD or ChatGPT? I use ChatGPT. I don't even know BARD. Yeah. Exactly. Exactly. That's the whole point. You know what I mean? So it kind of gives you a feel of like I think it's a great thing there is competition in AI because it allows, you know, for innovation to happen.

Um, but the overall point is back to the core of the question. I think it was around 20 2018, 2019, which was like the first time we started hearing um generative AI music using transformer models. Um, like how would you begin? How would you actually use it at that point? Like what would you do? Like you'd go onto a what, like a website Honestly. or like a what would you actually do? And what would you get? I would argue, I would argue it was unusable. Like no, no, seriously, if you were trying to make music, like as an artist, it was unusable because it would generate a MIDI pattern. Um, so the whole idea of artificial intelligence is broken down in the sense called machine learning, right? And machine learning is three steps it's the training data, your algorithm, that like the actual ML model, the algorithm, and then um using that model to make predictions about the future. So that's the basis of all we're talking about with generative AI. It's actually machine learning. So to that point, all the models that were available, most of them were pre-trained models. They were like already trained.

So the only thing you can do is that generate step. So when you generate based off of a model that someone had created off of some probably random open data set that wasn't high quality music, it made for the models to not be high quality in themselves. You know what I mean? So yeah, like garbage in, garbage out, garbage out. That's exactly what it is, it's garbage in, garbage, out. garbage out. That's the whole concept of AI for sure. Yeah. Um, but yeah, so so back then I would say um, so that same magenta project, you can either um use their model they already trained, or you can train your own. So when I trained my own was when I started like my the light bulb started to go off, like, oh, the biggest problem here isn't even necessarily the ML model, it's the data set. It's the input. So if you can exactly. And so if you can increase the quality of the input, I wonder if it would actually sound decent. And there's there's this whole article I wrote about like the breakdown process of like retraining a magenta model with my own data. And honestly, I I grew up a rapper and making that AI model was the only reason I was ever able to produce my own beat.

Like it was actually pretty fun, because like I never knew music theory, but at least I knew based on my historical like uh friends and producers I worked with, if I fed this to an AI model, when it generates things, it will be like in a certain key and scale, so I'll feel confident that like it's like correct. And then from there, all I would have to do was select like the sounds, you know, like use my VSTs or whatever. And that's more like an of an intuitive process rather than like picking where MIDI should go or whatnot, you know what I mean? So is it safe to say that lemonade was created because a rapper had to fill a void of not having a beatmaker? Straight up. That's exactly right. That's insane. I'm not even kidding. That's exactly right. Wow. Wow. Can't get these can't get these humans to do shit. Might as well make these robots work for me. No, but but okay, so that that's a really good thank you for that, that kind of background on where we've come from. Now I want to move into kind of like where where are we at today? Meaning, what are the most common use cases for musicians using artificial intelligence, right?

So where are we at? Yeah, totally. I mean, I think for one, the good news is I think the full song generation market is still something that's being figured out. And when I say the good news, like it's not like generating full music that can replace human musicians. And that's the good news right now. Um there are companies out there like Suno and Udio that I know can do like you know, three minute song generations or whatnot, but for the most part, it's not anything that you would actually want to listen to, like on Spotify, and like actually be like, Oh, I want to subscribe to this like artist that kind of sounds like Drake that's AI generated, like that doesn't exist as of October or whatever day today is 2020. But ironically, like I've seen some fully generated songs in like just kind of meme posts, and they're kind of like perfect for like ironic listening sometimes. I gotta admit. Well, that's a good point. I feel like there's probably a new industry that might be opened up because of AI within music that's like completely different than what we're used to. And it's that concept of like meme music is what I call it.

It's kind of just like, yeah, just like that song you made with your friend or whatnot, like humming during the weekend, and then you gave it to Suno or Udio. And I feel like that might be like a cool, like inside joke thing that has the chance to go viral. But at least for me, like at least the music that I love listening to, just like at the gym or like you know, on a drive, even though I never drive in in New York City, you know what I mean. I was gonna say you have a car. Yeah, yeah. How fucking rich is this guy bro? in New York? I took I took a pay cut, a big pay cut to build lemonade, actually. But I think it's not a hero. Oh, I know. The startup life, you ain't having a car in New York City, bro. No, no, no. But but you know what I mean. It's like that when you want the emotional connection with music, I feel like that's still missing in the full song generation ecosystem right now. So, so yeah, so that's one side of it. It's like there are full song generators. Um you can do it, you can do it today, though. You can do it. Okay, sure. Meme music is a good way to put it. I'm like if you want like that like happy birthday custom song that's created or whatnot, Yeah. That's like got your name and all the bells and whistles. That's there.

But it's the other side of AI creative tools that are assisting the music producer that I'm like way more interested in, or just assisting the creator. Yeah. Um, and like, you know, lemonade is the company I'll talk about, because obviously it's my company, but sure it's a good example of um what a few other like companies are doing right now, which is number one, trying to figure out how to do things ethically where like every single line in our algorithm in terms of how we trained our models, if you go back to that machine learning three steps, um we've gotten consent and met with each artist that we trained the algorithm on in terms of like every single line of data is like artists got paid for it. So that's one thing. Number two, um it's all about assistive generation. So like we generate uh eight bar audio samples and MIDI samples, right? And you guys are very familiar with what those are but for a producer to have infinite access to generate in a key and scale or a mood and there's going to be even more generation type soon where they can feel like this sample is mine. Like mathematically it's quite unique compared to like using the top splice sample for example and knowing that um there's gonna be no copyright issues or no like weird some kid opened like uploaded a like uh royalty free pack that was stolen off of like Kavi's site or something of that nature you know so all of those things are completely like you know solved by using generative AI at least the lemonade approach because it's from data that has been um ethically sourced and we have algorithms in place to try to make sure that we never output something that's similar to the training data like the exact same thing as the training data.

So what I'm getting to is um it's really assistive tooling. So if you've got writer's block or you're in the studio and like you're with a rapper for example like I've seen this actually play out firsthand Or, like you know I was actually in the studio with Don Cannon and Don Natalya, and I think I was telling you guys about this. And it was the coolest feeling, just like going through melodies together while you're in an actual music studio and having the entire room be like that's the one, then cook up with it in that exact moment. You know what I mean? So it brings this element of like an infinite piano player with you in the room. But again, the whole concept of like trying to make sure the artist makes the final decisions of like you still gotta drag it into your DAW. you still gotta throw your drums on it, though choose the arrangement, you know, choose your sounds. List goes on. But yeah, we made very intentional decisions not to be doing like full song generation or trying to even attempt to replace the producer in any way. So were you playing lemonade generated samples for Cannon? Mm-hmm.

Yeah, that's what he's saying. That's fire. They picked one and cooked up with it. Yeah, yeah. He fucked up, ate Something crazy. I think there's a snippet on YouTube. There's definitely snippet on Instagram on the tomorrow profile. Okay, we need that team. But yeah, yeah. I'm curious. I'm curious how um, I'm sure you've thought about this, but I use Logic Pro. Yeah, me too. that's where I make beats and I cook, yep, love logic. But um, and they the in a recent update, and I think actually a lot of Daws are doing it now, they're introducing their own kind of generative um I guess it is AI, yeah, generative AI. Like composers or whatnot. Uh-huh. Inside the the DAWs, they've had like the drummer one for a while that I never really liked, but this new one uh that does MIDI and it is like pretty good, you know. So, how how does a company like yours stand out from just if it's built into my program, you know? Yeah, totally. I mean, I think first and foremost, I try to stay as focused as possible because like there's so much noise happening in the AI music space right now that like if we paid attention to everything going on, I feel like I would have shut down Lemonade like months ago, you know, because there's literally a new company being born every night. So the first thing is just straight staying focused on like the core competencies and core values that we like believe in, which is really high quality algorithms and continuing to support our existing customers, getting feedback from them and iterating based on that feedback.

So, like that's the first, like overall, just like stay focused, even in the midst of other people building. Especially because exactly kind of like you said about like the drum um maybe uh the drum machine and logic, you don't use that for some reason. No, they they've had it had that for a while, and I think that's based on you know some of the older, maybe more magenta-esque type of models. Gar was garbage in, garbage out. Yeah, yeah. But but the I guess that whole concept and idea, I think scales to when DAWs also take on these types of projects where like what is the DAW do really well? It's recording and helping you produce and things of that nature. But as they add features, like like there's a reason they're stock plugins, and then there's a reason why you use like scorch or like you know, uh UAD plugins here or whatever. Exactly. There's a level of quality that it comes with. So I think that's the pocket we really want to like double down on. And I think it's a great sign that DAWs are actually implementing it because clearly they've gotten feedback going to back to my first point from their customers, they need help with inspiration. And the hope is, you know, they'll use a tool like that.

Maybe they'll get hooked on on the idea. They want more variety or they want to go even deeper. And what's dope is with Lemonade right now, we have a seeds algorithm, uh, which is 100% royalty-free. But we also have a product called Collab Club, which we recently launched. I don't know if you guys um caught that. But we've interviewed Kato on the track, and I'm also like a I'm also a Pain one fan. So like love both of those guys. Yeah, yeah. Those are super cool dudes, like every single person. So we got Kavi, Kato on the track, um uh DJ Pain One one and Mantra. So all of those guys are like super cool, super talented. Took us a very long time to work with them to train their algorithms. But the point I'm getting to is like that logic in MIDI generator is not gonna make trap style melodies like the Kato AI model is going to. You know what I mean? And like we can only scale from here, you know. We we've got a decent amount of like input from artists who want to join the collab club, which we're hyped about, some big names that you know, I don't want to put them on blast right now, but like super excited to see the interest because it it that's but probably been a cool part of even launching Collab Club is seeing the appetite for artists that are like, Oh, I'm down to work with y'all if I can train my own model. It's like whenever you can train your own model, people seem way more open to it than if you're using a random, what they call black box of like you have no idea what it's generating, who it's from, or whatever.

But when you have like a Kato on the track face to it, one, I think humans connect with it more and they know what they're getting, but then two, like bigger artists are like, oh shoot, like like literally people a year ago that like I'm not gonna name names that would be like I would never rock with lemonade, and then be like, oh, I can have my own lemonade, I'm definitely down to rock with y'all, you know. It's just interesting to see that psychological switch. They always come back around, my friend. Yeah, trust me. it's not even that, but it's a cool thing because obviously artists should be protective of what they do. But if you can put yourself into the machine and have the machine kind of sound like you and keep it private, because that's one thing we've been considering is like, should we allow artists to train their own models and even have their own private environment? Um, that's where I think people will like really enjoy AI because like the whole idea of it's cheating kind of like feels even less like that because you're actually just collaborating with yourself if it's your own model, you know what I mean. And I and I don't think AI is cheating to be clear, and we can talk about that.

Oh, I I've always hated the whole cheating argument. It's like people thought that like when plugins were created, like, oh, you have a digital version of this $10,000 box that people buy. Like that's cheating. Cheating. What do you mean? You have a mic that makes you sound better. Like cheating. That's cheating. Cheating. It's like, let's just go back to you know, banging pots and pans together. What are we doing? For sure. And it's cool because um, so I have a um podcast with the tomorrow team. And on one of the shows, uh, OG Parker came on, and um we had uh I think Canon asked, or I think I asked OG Parker, I was like, yo, do you think it's different? Like, do you think it's cheating to use a four-bar generated loop? And he goes off this whole rant about like it's no different than using Splice, arcade, and all these tools we're already used to. Yep. Um, so anytime there's new technology, there's gonna be friction, there's gonna be like an adoption curve. But unfortunately, like I think the winners are the people that are willing to be like, you know, swallow their ego and be like, let me actually learn how this technology works and use it, and all of a sudden it's their superpower, their competitive edge, you know. Yeah, I'm not gonna use a tractor.

We gotta farm manually. Yeah. The hell is this? You're right, right, right. Exactly. Boycott computers. Um, no, so I do want to, I think this is a perfect segue. What is the general attitude towards AI within the music community, right? You probably hear it most. What are people's opinions? Are you divided? Yeah, like what do people think about it, right? It's crazy because I'm so grateful I grew up. So I actually grew up a rapper, believe it or not. It's like my first and first and foremost, my love. And I I actually appreciate the fact that you put MJX in the calendar invite. I don't know if you realize this, but that is my rapper name. Okay, fire. I didn't know where the X came from, but I was like, that's kind of hard. So yeah, yeah. No, that's great. I was like, oh shoot, they actually did their research. But but but anyways, I had no idea who was a rapper. Just for the claim it, bro. Own it. Bro, I listened to your first song. Hey, listen. That first mixtape, bro, it's fire. I still have my first song. That's that's what I'm grateful for with music. Cause like when you're creative, you record everything. Yeah. So I still have my first video I did.

It was actually in like this basement of the section at home that my family used to live in. And I was like 11 years old, rapping with so much energy and passion. And I was like hyped. I was like, dang, good for you. You know, life is not going great for you in that moment. Like, you're like staying positive and dedicated to, like, you know, chasing dreams that are bigger than money is cool. Being a rapper young taught me like when I would upload stuff on YouTube, the amount of hate that I would get, like it really was 5050. It was like someone was like, dude, keep rapping. You're gonna go somewhere. And I'm like, all right, let's go. The other half's like, you fat Indian kid stop rapping. And I'm not even kidding. I'm like, I I was way skinnier than I even am right now. So it was I was like, why don't even understand this comment. But like obviously as a kid. I was gonna say, I was like, are you fat? Like if you're on a video screen right now. Yeah, yeah, yeah. Look like you're in good shape. You don't really, yeah, I appreciate it, guys. Yeah, you look great, man. Four times a week going to the gym. There you go. Fuck that commenter, bro. Seriously.

No, no, but but but my point is it helped develop like a thick skin as a young kid. Um, and then you know, I ended up getting deep into technology simply because my family grew up broke. Um, and I was like, I gotta figure out how to start making money, support the family, etc. ended up working at Google for five years, actually, and that was a dope opportunity, you know, nothing but love um for that that situation. And by the way, I found you uh over a year ago. I love the content about hey, I quit my job at Google to create this tool to assist artists. Like that was some of my favorite. I saw that on TikTok like forever ago, man. So yeah, it's crazy because that's actually where things popped off. It was all on TikTok, which... there it is. For the artist, for the artist story, that's true, but it's also true for the entrepreneur story of like if you've got a small business or an app or whatever, post it on TikTok, and like it's crazy how many sales can come in just off of like one of those videos. So then launching that, so go going back to your point, TikTok videos back then, Instagram videos back then, right when I launched Lemonade 1.0, um, which was I think two years ago, the it was super negative. I would say 90% negative, 10% positive.

But but so what were the negative people saying? Like, well, what were the comments? It sounds like shit. Well, hold on. Let's very similar, yeah, very similar to what we've been saying so far, which is like one, it's cheating. Like, you know, this is whack, this is cheating. Number two, there were some really harsh comments, like, you know, AI is like a cancer, and if you're building AI tools, you're like creating cancer for musicians. Jesus, wow. Um, yeah, so it's very emotional. You're the devil. And then the other thing is to your other point, it the quality was not as good. So in 2019 or 2018, in France and twinkle twinkle little star, let's say in like 2021, we were hitting like I don't know, um, that song Canon by Beethoven. Is that by Beethoven? You know the one? Yeah, canon number one. that's actually a terrible example. That that's a that's a banger. I was like I was like, wait, you mean one of the most highly regarded classical songs on planet Earth? Yeah, let me take a step back. We're still not there.

Uh, but you know, you know what I mean. If Twinkle Twinkle Little Star was a one out of ten, hot cross buns. That's a good one. Okay, that's much better. Um, but then you know, so people had comments about the quality, and like like jokes aside, it wasn't like hot cross buns. You can get some coherent melodies and chords, but there would be a reason why you wouldn't use this tool over like something like splice, for example. Yeah, um, but as the quality started to get better, the naysayers definitely decreased, but I would say it was still at like a 70 30. Yeah. Um, but it was until we released Collab Club that the haters, the the like persona of the haters decreased dramatically. Like I would say it went 70 30 in the opposite direction. Like when you have people like Kato on the track actually going out on the limb to say, well, this is my AI model. It can help you create music and it's not replacing you as an artist. And to be an artist saying that, that is the ultimate like way to like evangelize like AI adoption.

It's like an artist actually sharing, a successful artist sharing that they're open to it. Um yeah, you had a champion at that point. It was it was like these people who are trusted by other creators to have the best interests of creators at heart, he's saying, why the fuck would y'all not give this a try as opposed to a faceless you know company that's worried about their bottom line? Yeah, exactly. Exactly. Because even like like so many uh producers or artists are like, yo, you gotta post content right now, you know what I mean? So if if someone that was a quote unquote uh a someone that you're not looking up to saying that, it's one thing. But if like you know, Kato's like, yo, start making content, it's up to you to decide if you want to make it or not. But like it's an inherent truth. Like if you want to start making content to blow up, there's a you gotta put in work to make it happen. So yeah, when when artists are like talking about like the AI part, I think it has changed the narrative some, but not only that, um it's also um let's see what I was gonna say. Oh yeah, man, working with these artists, it was a crazy long process and journey.

Like, I feel like we all got scarred together building these models because like it wasn't like you know Kato sent me a catalog or like I use Kavi for example. It wasn't like Kavi sent me his catalog and like you know we trained these algorithms and it was done. Like Kavi had like high standards for his model. So like in the first time we trained it it took you know you feed it to this AI algorithm it takes like two days for it to learn the patterns and then we generate a bunch of stuff with Kavi and he's like this is like not even close to like what I would actually want and he would give a bunch of qualitative feedback that we would then add to like the scientific loop for retraining the algorithm but from start to finish it was like a six month process. Damn. So I guess my point is it's not even that these guys are evangelizing about it. I think there's just a certain level of like when you put in the work you can kind of stand behind it in a way of like this isn't a gimmicky like we're trying to like get a bunch of sales and like leave as a company like we're genuinely trying to build and invest in like the future of how do you build scalable business models with the artist while advancing AI. Yeah like it wasn't a a quick money grab where you just like spit out some shit and see how many sales you could get oh my gosh We spent over a hundred thousand dollars on just cloud spend, yeah, just to train algorithms. I'm already knowing, bro. So to be clear, are like, you're telling me a nine dollar subscription like we are like you know upcharging or whatever like the amount of like love care and just like money that we've already invested in building these algorithms is it's crazy.

But but yeah you know and when you're invested in more of the the long term, if you see if you see this journey on the X axis, like if you if you're more invested in like you know, the five plus year journey, investing right now is all that matters. Like building goodwill with artists, you know, building technology the right way, doing it ethically, because you know, there's other companies out there that are like saying they're using AI, but it's like just like a MIDI pack in the background that's like not even generating AI and using as a marketing term. So there's there's a lot of like money grabbing situations right now, but kind of like I said earlier, if you just like stick to the simple stuff, focus on your vision, build with customer feedback and like do things the right way. The hope is five plus years from now, artists will be thanking people like us for thinking about that future instead of being like, oh, now there's no company that's trying to think of a compensation model for AI when and if AI hits, you know, um, production at scale. Hey, Henry here.

Sorry to interrupt this podcast, but there is something I have to address. As many of you know here at one more time, we love doing live music reviews. We love hearing the music from our community and providing helpful feedback that is hopefully helping you further your career. But if there's one piece of advice that we are tired of giving, it's that your master is too quiet. We shouldn't have to turn up the volume for your song just to have to turn it back down for the next song. That's not good. But Henry, good mastering engineers are so hard to find and so expensive. Enter DistroKid's newest tool called Mixia. I master music myself, and while I don't think AI mastering will ever replace the quality you get from an experienced human mastering engineer, it's pretty damn good. And if you already distribute through DistroKid, you can now master unlimited tracks using Mixia. And if you don't, we have a link right here. You can sign up for DistroKid with 30% off, upload unlimited music yearly for an insanely cheap price. Humans perceive louder as better. It's dumb, but it's true. Use Mixia for a clean master, and you won't sound weak on that playlist. Let's get back to the episode.

Obviously, we know the naysayers and some of the things that they have, you know, complained about or commented about. What you know, what have some of the most positive comments been, right? Like, why do people love AI when it comes to making music, right? Like, just so it might, and some of them might be obvious, but like assume that they're not, right? Like, why should an artist or a producer, you know, consider using AI in their mix, right? Yeah, for sure. Um, it all depends on the tool you're talking about here. Like, for example, I know for like audio engineering, it's crazy. Like, shout out to so many cool companies where like when I just want to do a quick voiceover, and like I'm like, I've got like three laptops at home, right? I got one Mac with like my entire vocal chain and all my plugins, and I got another Mac that has none of that. So anytime I would have to like get my vocals right, I would have to like send my vocals from this Mac to my other and throw all the vocal chain and all these things. But then there's this sick AI company that does audio engineering, and like I think it's not Landr, Landr is a good example of one too, but there's another one that's called I think it's like um Vox or like Voice X or something like that.

I'll have to remember the name later. But you push one button and it gets your vocals like 90% there. So like increasing the workflow in ways that AI can't, I think it's an it's an incredible thing. So personally, I've loved that. But in terms of customers of like what they've liked about Lemonade, um, one is just how quick it is and like prototyping ideas. Like, I think we saw stats on our back end where like it takes two button pushes for someone to drag a MIDI file into their DAW, and a button push responds within 500 milliseconds. So the fact that people on average are just pushing it twice and cooking up immediately, like that's a like pretty cool signal for us. Um we're kind of like Hinge in a way. If you guys know that dating app where lemonade's on a mission to have you close lemonade as quickly as possible so you can get back to cooking up. So one, like the speed of inspiration, like the custom fit inspiration of like, you know, if you want something in that specific key and scale, like it's endless in terms of what you can get. And because it's a MIDI and audio samples, you have the ability to drag that in and like fully customize it and things of that nature.

So yeah, a custom fit for the project. Um let me try to think if there's anything else that's like major. Yeah, I think the productivity. Oh, sorry, go ahead. Yeah, here's a big one. I think it's the um uh educational piece that like indirectly happens where when you see everything labeled very like appropriately within Lemonade I'm like, okay, I want to there's this thing we call it inspire me mode, but it still tells you the key and scale, and it makes sure the MIDI is perfectly in that, right? So when you drag it to your DAW, um, there's almost this like feeling of like, oh, I know this is in C major. So I've had many like customers, including myself, be like, when I'm in my DAW, I can now force correct my DAW to be in C major, and all of a sudden they start learning what is an actual key and scale or whatnot. So we we have had customers that reach that and be like, yo, you've kind of reverse taught us how music theory works because I've used the tool enough, I kind of know where the black and white keys and things like that are for a certain key and scale. Yeah.

So that's been a fun thing we want to consider doubling down in. But here's the thing that is gonna be crazy. Our team is working on right now. There's two things. Uh, y'all know Sauceware audio. I don't think so. I don't think so. Okay, there's a producer named my best friend Jacob, super different dude, been a great friend of mine for like two years now. Um, anyways, he runs a company called Sauceware Audio. They have the craziest sound VSTs. Like they've released uh, I think it was um so they have Scorch a few years ago, Trifecta last year, and now on November 1st, we're co-releasing our first product get together called Spawn. That has been crazy because Spawn is taking the best of the lemonade MIDI engine with the best of a human created sound engine, and like intentionally pairing like the melody layer, the chord layer, and the bass layer um with the certain sounds and MIDI, and it's like full customizations. Um, anyways, so one that's in the pipeline coming November 1st. But here's the thing where I feel like in the creation world or did the generation world that people haven't cracked the code on well, and we're hoping to this quarter.

Um, it's this concept called conditional generation, right? Like it's one thing to um generate based on tags, but imagine you play a melody and AI immediately plays you counter melodies, B sections, etc. That's what's in the pipeline right now. Our science team is cracking the code on it like as we speak. Um, but we're hoping to make some progress in like November and probably you know have some early prototyping things that uh that would be fun to jam out with. That's gonna be fire. You're basically gonna you're basically gonna have, like, a built-in co-producer, yeah, you know, who's just mirroring your energy and, like, mirroring your, you're automatically creating a tiny model based on your input of your melody, and it's spitting back kind of based on that. Um I'm curious too. Do you have anything um that like is AI gonna be try to like match because some of my favorite things about like playing things myself is the little like human elements, like velocities being kind of like a lighter touch on some, or like even maybe like a wrong note occasionally, or like yeah, it's just the little quirks. It's not so perfect, yeah.

Yeah, that's something that's good to demystify. Like AI can already do that. And the reason why AI can is because people are now learning, going back to that three-step process, it's all about the training data. So in the past, the training data was a lot more robotic because it probably wasn't real musicians that was training the algorithms. It was like engineers and scientists that are like, what's this AI music thing that I can use to generate? And like grabbing whatever you know, hot cross buns data set they can find. Shout out hot cross buns, dude. It's made its way into the podcast twice. Yeah, yeah. But now when you work with like Kavi, who's got like dope strumming on his chords and like intentional velocity that's like way like lower at times and like way way harder at times, etc. The AI can learn those patterns. And if you even mess around with lemonade, and again, I'm I'm shouting out lemonade, but there's many other AI companies that do this. But if you pay attention to the MIDI, it's not perfectly on the grid. Like it is actually like you'll see the strumming slightly off, you'll see the velocity slightly off.

Like the times is perfectly on, but but it's dope because mantra, for example, who's more of a pop style producer, he does a lot. He's a super dope dude and amazing producer, but some of his like pop melodies might be more on grid. So it's cool to even see the difference between the Kavi model and the mantra model, and the AI just learned from his data set, kind of matches it, yeah. In a DAW, it might be like the a knob that says humanize, and maybe you can like add more, you know. Exactly. But but the point of this is like out of the box, the AI is humanized because it's learned on the human data. Um it's a funny like like um kind of headline thing to say, but like I I genuinely believe this. Like AI is actually dumb. Like it's not actually intelligent, it's the human that makes it intelligent. And I just posted like a little piece of content about this recently. But for example, imagine you're trying to train a melody algorithm that was happy, and you put a bunch of sad songs in a folder, and then you call that folder happy, and then you give that algorithm to someone else and say, yo, can you make me a happy song? It's just gonna output like super sad songs, you know, and like the the user's gonna be like, What the heck?

Like, this is like not like this is a dumb AI. It's actually the human that literally just feeds it what it needs and labels it appropriately. So AI is nothing without the human actually taking the intervention to train it. So technically, AI is only as smart as we are right now. Um, so if you're ever frustrated with an algorithm, it's because the person who created it or the person who trained it probably didn't take in all the different biases and things of that nature. Oh, damn. It's a good point. People are like, this uh algorithm, it's like a dude made it. Yeah, yeah, exactly. Blame Henry. Blame MJ if lemonade sucks. Yeah, we got a team, man. We got a team. All right, blame his team. Blame the team. Yeah, but I'm kidding. I'm kidding. You all know the buck stops with you, my friend. Um, so but yeah, do you know too much about the vocaloid software like space, or is that not really something that you've de like done too much? I've messed with Vocaloid AI a ton. Okay, what do you mean by vocaloid? I don't know.

That's okay. It's like something I heard. That's why I was hoping you would just be like, oh yeah, Vocaloid software is this. thing. What I'll say is vocal, because terms can throw off everything, you know what I mean? Like vocal AI is a good thing. Yeah, what is it? It's the same thing. Sure. Okay, okay, gotcha. Either you either you uh record an audio and you feed it into it and it changes it into another voice, or you just type lyrics and it spits it out based on a model that that already exists. Well, yeah, those are two. Now you're the expert AI model. I was asking you for exactly. I was asking you for help, bro. Yeah, I mean, I I I kind of get it. We did a whole episode on this. But I was I was saying, I don't know how this works. So is that how it works? Like, how do how do like AI voices come to be? Yeah, there's two different types, and you broke that down perfectly. Like one of them is called SVC, which is singer's voice conversion. And that is the one where like if I were to rap into a microphone, yeah, and then upload it to like some service or really just an AI model, um, and I would choose like the Drake voice, it would output like a Drake sounding voice. Okay.

So that is the SVC model. Sure. That one's actually pretty interesting in terms of like I've heard songs, like there's this one by by like the AI juice world voice model that is so good. And like the reason why it's so good is because that vocal model is gonna be garbage if you're uh it's the same thing with your input is garbage. Garbage and garbage out. Exactly. It's like auto-tune, right? Auto-tune can only take you so far, even with SVC, if you don't have the proper like key and pitch or like cadence, it's not gonna fix your rap. Like, you actually have to have like I guarantee that ghostwriter dude with heart on my sleeve, mm-hmm, yeah, yeah, yeah, yeah, yeah. He or someone, whoever rapped that, if you took off the Drake voice, it would still be hard. Yeah, you know what I mean? Like it would still be so good. The only thing that happened is like there was a Drake masked voice over it. Um, but the key is you still need a talent to create it. So SVC, I'm actually pretty like bullish on in terms of like the good it could help artists with.

Um, especially if like you don't like your own voice, and that's been a reason why you haven't made music before, or even for someone like Ghost Rider, because to my understanding, he's more of a producer. So to help a producer make a full song on their own, I think it's just an amazing thing. So that's the SVC side. But let me just be clear, everything that I'm about is doing it ethically. So if you work with an artist to train a voice algorithm that's that they can lease out, then it's dope. But if you're just random or using random technology to like, you know, copy people's voices, then I'm just not about it. Cause just imagine if it happened to yourself, you know. Um but on the other side, it's more like um text-to-speech kind of AI voice generation. That one I'm not as familiar with. Um, but the whole comp regardless of being familiar with it or not, the whole concept of that machine learning thing is still the same of like training data, model, and outputs. So probably trained on a bunch of um, you know, vocal a cappellas at a certain cadence, maybe a certain key and scale, you know, list goes on. And then based on that, when you use the text to generate, it's probably just picking one of those cadences and one of those like vocal like timbres and then just generating something from there.

I think that one has a That's a real assumption. That one has a long way to go, I think. Yeah, what's the point? Yeah, yeah, exactly. I agree. The SVC is dope. I think that's actually pretty good right now, especially if you like drag a output from another vocal model that you've used, throw it into your DAW and actually mix and master with it. Yeah, that's when it's like you can barely even tell at that point, you know. Yeah, so so I I want to touch on like the revenue stream opportunities, right? Because for me, I look at this from not only a creativity perspective, how can you be more productive, right? How can you be more inspired? All the things that you talked about as some of the like positive benefits. I'm wondering what your thoughts are around just like the business of AI for a creative, right? Because we're big on helping creatives turn their like passion for creating into money, right? Like, we want you to be able to make money and like have a like make a living off of your art, right? Whether it's AI generated or fucking human generated, we don't necessarily care at this point in 2024. But I'm curious as to like how you see revenue streams being unlocked with AI.

Yeah, it's it's a lane of opportunity for everyone right now. Like if you're a tech company or a tech person, if you're a musician or yeah, if you're a brand new market, which is like a tech musician, I guess those that's a new persona being created. But yeah, there's a lot of opportunity to make money right now in AI if you're an artist. Like, number one, Lemonade pays out thousands of dollars per month to artists just for contributing their data, and like again, that's by doing it the ethical way of like we work with artists and you know, we make sure their data set's clean, we pay them for it on a monthly basis, and we don't take away any rights from them on that music, so they can keep selling it on BeatStars or whatever they want to do with it, but we're just saying, yo, that melody you created, this is another form of passive income. So that's one thing, like straight up, like uh monthly checks are being written, not just from companies like Lemonade, there are several. Um, but the second thing is that, and this is something that is more unique to Lemonade, is like we're still figuring out like if we want to open up the keys to all of Collab Club to have anyone train their own AI model, but you can almost imagine eventually we're gonna get to a point where you can imagine a marketplace of models where like you want to generate a chill melody. Well, the chill melody came from a certain producer's model, and as a result, they're getting when you drag that into your DAW, some type of compensation.

Yeah. And on top of that, with Collab Club, we actually make sure the artists are protected where they they can set their own terms in the publishing. So if you use a Kavi model and you get like a major placement or whatnot, Kavi technically is entitled to 5% of the publishing. So if you get to work with some of these AI companies early enough, especially because like right now it's still being figured out, it's the wild west. There's some crazy deals that can just be structured where you're not compromising yourself as an artist and you're making more money and you're still doing what you do, like what you love, which is making music. But the last thing I'll say is this is obvious for any business. So this is like one of those like don't leave the house without your phone kind of comments. It's like just be careful who you work with. Like, don't just give your data to any company that's an AI company, like learn about who they are. Like, have you seen their CEO ever speak? And if you haven't, it's probably not a good thing. You know what I mean? Like, like actually get to know the like the heart of that company, because on the other end, AI is a super powerful technology, and if the wrong hands get the right amount of data, then it could be pretty pretty bad in terms of the high quality algorithms that are cheap monthly subscriptions that are now making the beats that you sell even more commoditized. Yeah, it's interesting.

Like, you know, the 5% publishing example. I mean, if you were traditionally collaborating with a Kavi or a Kato, they're probably entitled to way more than 5%. So it's also like somewhat of a win-win, right? Like Kavi's not having to do shit, but he's getting a little bit of pub in case you know, songs start to take off, or like you know, you start getting a bunch of them and they they add up. And then same thing for the the other producer on the other side, it gives you access to once again, like high quality like samples and and you know, collaboration types of uh sounds, right? Where you're not having to shell out fucking 25% pub or whatever. But are you gonna go say I collaborated with mantra? Can can they can they put co-produced by my or by Kavi and Kato or no? Yeah, so how does that work? That's a good question, right? the term it's a great question. Um, it hasn't happened yet, so it'll be fun to see when it does because we just launched Collab Club in August. But the terms are when you break a million streams or when you land a major placement, you need it's enforced that you need to put them on the publishing, like whatever term they set.

But in terms of tagging them as a collaborator or whatnot, I think that's more to the artist like um kind of consent, or like you know how on Spotify you can tag whoever you want technically, or you used to be able to. Oh, I can put Drake is an artist on my next release, and they might not catch it. I mean The amount of songs that I was on, and I was like, what is this like like elementary school thing that like I'm tagged in? So technically speaking, I think a producer could do that, but it's the same thing as if they didn't use their model, right? So then it's up to mantra, Kato, or whoever to be like, I'm down to actually collaborate. Because what's cool is we did a beat competition with the models um on BeatStars. and by the way, like Beatstars as a strategic partner of ours, like so dope. Like I really appreciated like working with them, and the idea was you gotta cook up with one of the melodies from one of these four models. Um, so all that being said, it was cool because now Kato, DJ Payne One, um, you know, Kavi and Mantra, they're gonna be cooking up a song with the winner from their um AI model. That's cool.

So it'll be cool to see an actual artifact come out of it. Yeah, um, but yeah, in terms of the actual credit under a million streams, we just haven't figured out yet how that's gonna shake out. But yeah, it's like I'm totally about to cook up a beat with a Kato lemonade sample, and it's gonna be co-produced by Kato. But to his point, you could put co-produced by Kato like on anything, and it's the same thing. But the cool part is, like, I'm sure if you did do that with Kato and you sent it to him on Instagram or whatever, he'd be like, yo, it's if it's fire, let's do it. You know what I mean? That that is the cool part because now him too. Technically cook up with his model, yeah, which was trained on his catalog. So there was some truth to you using it's like a collaborative cookup. So then it's just that last step of like, are you gonna be bold enough to send the DM and actually so I know you I know Luminate is mainly focused on serving producers, right? But um, but as far as like the artist and the songwriter, once again, maybe just your general thoughts. Like, where do you see AI playing into like the lyric, like songwriting and the like more artist side of things?

I know that's not Luminate's wheelhouse, but just in general, you've probably thought about it or seen some cool companies doing certain things. Like, what you know, what are your thoughts there? Yeah, I think some cool things that like we've even thought about is like, is there a way that an artist can like upload their own vocals and have like a beat referenced with it? That would sound like dope, you know what I mean? Um, of like, let's say you recorded over like a YouTube instrumental that doesn't exist, like you can't find that producer, or like that there's no like beat sales link or whatever, you're like, dang it, I had a banger. Or it's sold like exclusively already sold exclusively there's so many, so many like reasons why it may not, you know, work out. But if you could upload your that song that you created and have like a reference of like 20 different styles like rock like afro beats like EDM whatever just things you would never do before I think that's a really cool thing because then it could push you as an artist to try out different type beats or different like styles you've never considered before so one just like collaboration of like when it's a finished song are there other versions of it you could consider um like I said before like if you were an artist doing a DIY like project like at home in your basement with a cheap microphone and have no plugins like when you have the ability to have like some of those like maybe like sixty dollar AI audio engineering plugins and you can actually make tracks that are like reasonable you know it's not gonna be like you know engineered by like Leslie for example but it's gonna be like engineered to the point where it's good enough to upload to the DSPs and like actually get it hit. Like I feel like there's some big songs today that are not even properly mass mixed and mastered and just like we listen to them constantly right? 100%. That's that's another category you don't know Leslie, right?

I do know Leslie Braithwaite oh that's impressive. I was about to ask him to get Leslie on the pod if he if he knows I mean he's just you keep you know I'm be like all right bro. No, we know who you know now. we know who you know now shout out to Cannon because you know meeting him was just like an honor first of all but secondly like he's just like brought so many people on the show and it's so fun working with him because he's like he just text someone and they show up the next day. Trust me but anyways Hollywood Cole you were supposed to make the Cannon episode happen already so I'm on your fucking ass Cole. oh man dang, all right, all right, yo guys, chill. I can't be Hit up everyone. Who else do you know? Accountability out here. Yeah, that's pretty much it, actually. Now that I think about it. fair enough. But even things like um, I don't really like the concept of lyric generation because I feel like the words you gotta say, like, come on, like you at least gotta think of those on your own. But if you like wrote a verse and you gave it to a large language model like ChatGPT or whatever, and just was like, hey, can you change up some words in here or like you know, restructure this in a way that I could try it again? I think that's kind of collaborative of like if you're just like, you know, this verse is it isn't isn't hitting, but and you want to change like one or two words or whatnot.

Or like you're stuck or you're stuck on the second, you're stuck on the second verse, you can't fucking think of what to say on the second verse. Plug the first verse in and the hook, give me a second verse, ChatGPT. Like, what's good? I mean, I don't know. No, but just at least as a jump off point, like it's you're hopefully you're always editing, you know, and going back later. But like everything I when I go to ChatGPT, I'm I'm not often using exactly what it gives me, but it gets my gears turning. Yeah, I just think to be honest, in in today's music industry, there's like very little excuse for beat block and writer's block in my mind. like there's just that's like us. We you know, we used to not be able to come up with content ideas because you know, we're content creators, that's what we do most of. And it's like, bro, like shut up, ask plug it to make videos. about. Plug something into ChatGPT, stop fucking around and get some ideas and like start trying, you know what I mean? So I don't know. You're Totally right. I mean, I think that's unfortunately true for everything right now. It's not just music, you know what I mean? Like, yeah, like I don't know, like even earlier today, I was working through a business model, and I was like, probably I probably can't hit anyone up in this exact moment because I want to answer this real quick.

So I just opened up ChatGPT and I just went back and forth I'm like, what do you think of this business model? It gave me a response. I'm like, well, what would a producer really think? I'm like pushing it, and it gave me so many different angles or whatnot to think through. So, anyways, I guess the overall point is I I love the idea of like it really is like a jumping off point. It's like an inspirational point to get started. It for sure works for music, but it also has worked for other things, and like, you know, whether it's content creation, like workout routines, or whatever it may be. Absolutely. MJX, we've entered a final segment of the podcast. Let's go. Everyone's favorite part. This is the rapid fire rampage. There's gonna be three parts. MJX, we're gonna start with some short answer questions. Let me get close. Yes, and this is where you focus. The everything's on the line here. Bro, have you been called MJX this many times in like the last several years at this point? No, but I'm about to start demanding everyone. Bro, he is MJX from now on.

Put some respect on that name. Yes. I'm not responding. I'm sorry, did you forget the X? Go fuck yourself. I'm MJX. That's exactly right. Yep. This is the short answer section. Ask some quick questions. Give me a short answer. Very simple. Let's get into it. Okay. Starting with on a scale of one to 10. How would you honestly rate Ben as an interviewer? Nine. Oh, baby. Put my shit in a generative AI model. Get yourself a I'm gonna get my own generative host in a little bit. God, if I could get rid of you, bro. I would. Where does the name lemonade come from? Just real quick. I was curious. Yeah, I was in the car with my wife, and I wanted to call it Beat Lab, and she was like, That is the wackiest name ever. Like, we gotta like come up with something better. So we went back and forth for like an hour, and she thinks she's the one that said the word lemonade. I'm pretty sure I did, but we said lemonade, and both of us were like, oh, that's refreshing. And then I was like, what if you put an AI in the middle?

Yeah. Because it also is an aid. I I just thought of the A, the I swear to God. The craziest part is what is the point of AI? It's refreshing ideas. Yeah. Whoa. It's very good. I think you should tell that story on TikTok. That's a good one. All right. That's good. That's even an AI right there. Uh huh. MJX, you must replace your hands with something other than hands. Shoot, that's tough. It's wild. Cold brews. Cold, definitely cold brews. Ooh, permanent. Like, yeah. Yeah, yeah. Just always sip in cold brews. I love it. Dude's a cold brew fiend, bro. Oh, yes. In five years' time, what percent of new music would you guess had AI involved in its production? Hmm. 25%. And how many years did you say? Five. Five. I feel like it's conservative. I was gonna go like 40, 50. Nah, bro. These artists are gonna cling on to this human five years? It's a long time.

Although the production side, 2029. He had so many good arguments on the production. So much music being made right now. I feel like 25% is still, like, music. That's true. It's still like 50,000 songs a day or something. You know what I mean? It's gonna be on everyone's phone. It's gonna be in the dog. It's probably more than 25%. But for short form answer is 25%. Okay, okay, for short form, for short form. Uh last one of the short answer. What is your favorite curse word? I don't curse. He doesn't curse. How did you not pick up on that the whole episode? I don't curse. I did, but I thought it would be still a funny question. Okay. But like, can you say like which, like what it starts with? Fudge. Can you just like just give us like what you would The thing is like I don't curse, but I say what the F a lot. So that's so it is. So it's giving me the hardest time about it. So the capital letter F is my favorite curse. word. There it is. Can I just I'm always curious though? Hold on. Can I just ask, like what, you know? Why not? Like what, you know. Yeah, I mean, I used to curse a lot when I was in middle school. And then long story short, my dad got locked up.

Um, and then when we would visit him in prison, there would just be like a lot of just like curse words and things going around, and like my mom just like put in a lot of emphasis to be like, make sure you don't end up in prison, essentially. Like just a simple statement like that. Cursing equals prison. I'm a young kid. Yeah, exactly. There was like this like extreme, like, oh shoot. Um, so one, there's that, but two, like, yeah, I think there's a lot of like darkness in the world, and I'm not associating cursing with darkness, but I'm just trying to be like if I can be someone that's influential, impactful, and do it without doing quote unquote things that are negatively associated. That's dope. I like it. Fucking A man. He's right. That's a good fucking point. That's a Great fucking point, bro. If anyone, if any of my high school friends see this, they would crack up. Uh I used to, I used to have this thing called PD cam in high school. Because I actually like, you know, I was like on class council and everything, and and I would literally say PD cam to my big groups of like friends, and PD cam stood for please don't curse around me. Wow.

PD cam. PD cam sounded creepy at first. Like this it's MJX again. PD cam. Yeah, damn it. But I can see why that would be a little weird. I love it. I love see it turned into a good story. Wow. Part two is the this or that. I'm gonna give you two choices. You just pick one. Okay. Starting with smooth melodies or knocking drums. Smooth melodies. Siri or Alexa? Siri. Discover new artists or binge your favorites. Binge my favorites. Ranch or blue cheese? Ranch. Oh, but recently, blue cheese is coming close, but it's still ranch. Making a comeback though. It's still ranch. Kendrick or Drake. I can't answer this one. I refuse to refuse. I knew we weren't gonna get an answer. Yeah, I'm sorry. Oh good. Sexy teacher or sexy nurse? My wife. Such a good guy.

It's it's October. Which one would you uh have her dress up as for Halloween? Okay, there you go. That basically answers the question. I love that's a fantastic reframe. That answers the question. Teacher would be cool. I like you too. I like it. Okay. Guitar or piano. Guitar. Meditate or skydive. Meditate. Two-minute song or six-minute song. That has so much context behind the answer. Nothing further is added. No further. information. I'll go two minute song. Two. Two. And finally, go back to being a full-time rapper or leave the tech industry forever. And you can stay in tech. the, the, um, the, the brain in me wants to get into the rap game full-time again because music really is my first love. But the heart in me is like, there's not enough people that care about artists that are building AI tools. So I have to be in this space.

Oh, there it is. Part two. Dedicated. Living his mission, bro. Dedicated. I love it. Did you know he was a rapper? Or did you just freestyle that one? I know you asked some things like part three. Moving on. This is the word association. He's like, damn, bro. He's out here improvising and shit. I like it. I love it. That's great. Come on, bro. I'm gonna say one word. I just want you to give me the first word right back off the top of your head. Here we go. This is it. All the marbles. Starting with music. Oh, what was the prompt? Prompt is great. We haven't had someone like this in a while. Yeah, yeah. But didn't comprehend what it is you were doing. Also, it's like an AI guy would say, like prompt. Like it's such a like what's the prompt. Well, what's the question? Well, could you just say it? No question. I say one word, and then you just give me back the first word off the top of your head. Okay, okay, I got you. It's all that. Yeah. So your first word is music. Life.

TikTok. Waste of time. Family. Everything. Optimus. That's me. You're the model for you were Elon's model for the uh for the robot. Optimist. Oh. Oh, optimist. Optimistic, like optimistic. Optimus. Optimus. Oh. The robot. Overpromising. Sushi. Solid. Needle. Needle. Needle. Gross. Password. Not saying. Almost got him. AI. Lemonade. War. Bad. Baby oil. Good.

Matrix. Neo. Lil. Baby. W. W. Lemon. A. Let's go. What a rampage. This is the AI. God. MJX. Hey. Don't forget. Don't forget the X. Don't forget the X. Please say the X. I'm responding from now on. Please say the X. As always, please like, comment, and subscribe. We appreciate y'all tuning in each and every week. We are here. Henry, until next time, what are we doing? Getting out of here. Getting out of here. Peace. Appreciate you guys. Peace.