Episode 95: How SongPot AI is Changing Music Discovery (And What It Means for Sync & Creators)

You can feel the right track in your bones, but finding it inside a giant catalogue can still be painfully slow. That gap between what we mean and what search engines can understand is where sync licensing briefs stall, temp tracks take over, and great back catalogue gets left behind. I'm joined by Tiangu Zhu, founder of Songpot, to unpack a simple but ambitious goal: building AI that truly understands music as a language, not just as metadata.

We talk through the real-world problems music supervisors and media teams face when words fail. Genre, mood and “danceability” are subjective, tagging is inconsistent, and a song rarely fits neatly into a few labels. Tiangu explains how AI music discovery can analyse audio itself to reveal “unspoken similarities”, helping libraries and rights holders improve music search, speed up clearance workflows, and deliver better matches for sync licensing. We also get into how Songpot can sit in the stack as a platform or an API for more tech-native companies.

Then we flip to the creator side. Tiangu makes a clear case for human-centred generative AI: not replacing artists, but acting like a new instrument for producers and musicians. From prototyping ideas faster to turning a hummed melody plus a style into an instant draft, the focus stays on helping creators translate what’s in their head into something they can actually hear, share, and refine.

If you care about music supervision, music libraries, catalogue value, music information retrieval, or practical AI tools for music production, this conversation will stretch your thinking. Subscribe for more, share this with a friend in sync or production, and leave a review if you want us to keep bringing you guests building the future of the music industry.

https://songpot.art

Transcript

Welcome And Guest Set-Up

SPEAKER_00 0:14

Hello, and welcome to the music business body with me, Johnny Amos, Podcast in Our Birmingham in England. I'm the author of the book, The Music Business for Music Creators. I'm a music creator myself. I'm a producer, I'm a consultant, I'm an artist manager and a senior lecturer in both music business and in music creation. Wherever you are and whatever you do, consider yourself welcome to this podcast and to a part of this community. My goal is simple. I'm here to try and educate and inspire music creators from all over the world in their quest to achieving their goals by gaining a greater understanding of the business of music. Okay, so in this week's episode, everybody, I am joined by a fascinating lady by the name of Tiengu Zhu, who is the owner and starter of a fantastically exciting company called Songpot. Now, let me just tell you a little bit about Songpot. I became aware um of what Tiengu does because I'd heard about her and I looked her up on LinkedIn because I'm a bit of a like LinkedIn stalker, right? In in the most innocent of ways, right? Because I'm always looking for exciting people that I believe are adding value uh to the music industries. And uh Tiengu is most definitely one of those people. She was one of those kind of child genius type people who just knew how to like, you know, build stuff at the age of seven with like C and all these different coding programming. She's won all sorts of different awards and written and composed for all sorts of different, you know, pieces of literature. Um and she has built a song pot, right, which basically does two different things. It analyses, uses AI to analyse music for libraries, for music usage, so it understands music in order to be able to kind of make suggestions for discovery, for sync licensing and various other things. And it also adds value to music creators, it does not replace music creators, not by any means. It's the polar opposite of the goal of what she's trying to build, right? But it's something that adds value to music creators and and they she explains it in fact brilliantly as a new instrument, right? So it might to me it felt like uh she's building two really different things. Um, and then in speaking with her, it's very clear to me that actually it's kind of the same thing. When we think about uh music being a beautiful universal language, and it is right, it travels all over the world to lots of different people, but that musical language, that universal language, is also a very, very complex one, and sometimes even talking from human to human in order to understand the need of what should be done in music can be lost in translation and uh it could be demoralizing, right? And so I think what Tiangu has built is something that can understand the nuances of how human cognitive behavior works when it comes to creating music. That's a lot of information. She explains it very well, right? So I'm gonna hand over to the interview. Take note because I think that this is a lady that is on the rise in the global music scene. She's building something that I think is going to change the way a lot of people work uh at media companies and also at sync agents and possibly even for music creators too. Here's the interview. See what you think. Uh, but anyway, um so uh Tianko, welcome to the music business buddy. Uh it's really good to have you here. How are you?

SPEAKER_02 3:42

I'm doing great, thank you. Thank you so much for inviting me to this podcast. I have been a huge fan of it, actually. I I listened to many, many episodes, like especially the ones on like mus music supervision. There was one, yeah, there was one um that's by the owner of Catalog. I remember Frederick. Oh, yeah, yeah, Frederick, yeah, yeah, yeah. Yeah, and uh yeah, and another one that just came out, like the um I forgot the name of my.

SPEAKER_00 4:12

Oh, right. I know there's so many of them. I do every Wednesday there's a new episode, and it's quite hard to keep up. And I and I'm generally about four weeks sort of ahead, if you like. And so, and so I think to myself, right, if somebody asks me about this week's episode, I think, right, which one went out this week? Because that would have been what I was doing about four weeks ago, you know.

SPEAKER_02 4:32

Ah, I see, I see.

SPEAKER_00 4:37

There's so much to keep up with, though, in this business, isn't there? And I I'm thrilled that you've come on because I am so fascinated by by what you're by by you, right? To be honest with you, you know, and by and by your work and by what you're building. I mean, I've got I've got so many questions for you, and I'm really excited about what you're doing. And I I should start by saying this as well, you know, congratulations, right? Because this is a really uh, you know, an exciting era of your life and of your career um with the design, you know. I mean, you're designing and launching SongPot. I mean, I think it's going to be massive. Um, for anybody that's not yet familiar with Songpot, could you give us a kind of overview of the goals?

SPEAKER_02 5:18

Yeah, the overall goals of Songpot is that we want to build uh AI that truly understands music, that uh can serve the music discovery. So, for example, you could search music by text description instead of just metadata, the title and artist, but also like audio and video so that it can truly understand the music content and help you to find what you actually need. And um, this is super important for the say sync and uh in general DSP and all this um um platforms and companies that own a huge catalog, and then um the other side of building AI that understands music is that it can really serve the creativity. So, say for musicians and producers, I like to think about think of AI as a sort of new instrument because it what it does literally is synthesizing sound. And so it is a synthesizer. And uh right now we uh we do have companies that work on, like say you you have a text prompt and then you generate a whole song, but I think that's not really the point of creativity, and I think um we could actually build AI that augments creativity by making it a new type of instrument, a soundmaking machine. And this is fascinating for me because in the end, the goal is to the problem that we are trying to solve is to how to translate uh the music in your mind to the music that's to your ear.

SPEAKER_00 7:02

Yeah. You see? Yeah, yeah.

SPEAKER_02 7:04

Either discovery or creativity, but it's the same thing. It's uh how to get that thing out from your mind.

Fixing Sync Search And Metadata

SPEAKER_00 7:10

Wow. Do you know what? Two major problems there, right? That you've you know you you're working on the solution for. Um, and it makes a lot of sense. It's very interesting. I mean, let's split them out into two different things, right? So the music kind of discovery on the library side and the sync side, and then the tools for creators. But it's interesting because the second part there, as a producer myself, the amount of times where an artist will say, I know what I want, I just I can't think of like how to explain it or how to articulate it or how to, you know, and uh and like the old way would be like, right, let's think about musical language, music, music theory, or or maybe even just some adjectives to help us to get there as well. And you know, so even that can be difficult sometimes, right? So it's like we need this kind of like almost kind of like biologically in tune thing between humans to make it work, and actually we don't always need that, like that can be simplified. I mean, I'm preaching to the conversity here, you know this, right? That's why you're building what you're building. So let's let let's come back to that bit. Let's go to the the library bit because you know I speak to I speak to a lot of people at a lot of different libraries, and and and and um, you know, and they're all trying to do their best to make their music as as discoverable as possible. And you know, they've figured out the ways in which they can license that, but actually linking uh, you know, people that say music users that say, I need this music, but like how do I get there, right? You you're kind of working to solve that problem. So, you know, clearance and and and speed are like you know high priorities for like media companies that want music for television, for example, who are looking to license music and they've got ways that they can license it. And there's you know, uh, the accuracy on metadata is important and all that kind of stuff. But it is that something that you noticed, right, that you are trying to cater for?

SPEAKER_02 9:03

Yes, exactly. This is what we are trying to cater for. And we talked to a lot of like music supervisors and also like um audio production, you know, and people working in this industry. And I realized there's so um so much trouble in finding what they need when they know already what they need. It's just the technology now, the systems, that's that's not clever enough, let's say, to allow them to find the things even if they they they already know. And um and with AI, the the cool, the cool thing about it is that it can really like understand um not just you know like how we how we previously deal with metadata. We are not constrained to that anymore. So even if you have a catalog of like say poor catalog um metadata curation because there's something missing or the tagging isn't so uh correct or something, uh AI can now just describe everything for you so that you don't miss that. And also to find um something adjacent to it, right? Because sometimes you you try to categorize as into a few words, but it's so hard to actually categorize a song into just a few words, and you're you know, and um, and also there are so many words that can get your music say lost in the sea of music. Um right, um yeah, like uh say you you publish something and uh the you you you have your subgenre in indie rock, but there are so many indie rock there. And and then you you maybe you have like three words to describe the emotion or something, but but this is so limiting, and uh actually we we should have a better system for this.

SPEAKER_00 11:05

Yeah, that's a good point. Okay, so let's split that into like the two categories that that that I can think of that I can understand, right? So one which is like the metadata, right? Sort of creative metadata, genre, secondary genre, stuff like that, maybe stuff around mood. And then if we've got that there as text, and then over here, there's technology now that like the DSPs use, for example, to sort of do raw file analysis, um which which kind of like we think of it as like, oh, it listens to songs, but I think really it kind of listens to elements of songs and then puts them into pattern recognition to be able to draw links between how danceable something is or uh or how uh you know how how much speechiness is into it, you know, that kind of stuff, right? Um and so if we draw a link between the first part and the second part, i.e., imagine I'm putting a song now through a distributor onto Spotify, for example, and I want to pitch it to say some sync agents as well, then I might call it, you know, indie rock, for example, right? Yeah, um, but then really by the time it's analyzed, um, it might be like, yeah, it's kind of not really indie rock, it's kind of more like uh pop electro kind of synthpop uh with sort of guitars in there. So what I said it was, and what people, what the machinery thinks it is, is actually like a bit of a disalignment. Now, if the libraries are just built on the first one, then presumably music users go to it and go, Oh yeah, I want something that's like this, and then they type it in and they go, No, that doesn't sound like that. Like, no, you know, exactly.

SPEAKER_02 12:44

So you and uh everyone perceives like all those words so differently because all those words are so subjective. Um, say danceability. I think something danceable for one is not danceable for another, right?

SPEAKER_00 12:59

Yeah, yeah, yeah, exactly. Yeah, and that that that can be a problem in itself, actually, can't it? Um, so is the goal for songpop when it comes to this particular aspect of what you're building, is the goal for this to act as a sort of filter between music users and music libraries? Will you sit in the middle of what they're doing or or whereabouts in the market do you sit there?

SPEAKER_02 13:24

It can be applied for both. This is my answer, because uh it's all linked to music discovery for sync or for DSPs. Like uh in the end, you are trying to find something that suits what the user wants in a huge library. Yeah, and yeah, and and the technology that we are building is exactly the one that's trying to solve this miscommunication problem, let's say, in the current uh systems. And um especially, for example, sometimes you you could describe, of course, with words and they can be correct, but um in music there they're also unspoken. There are things that you cannot put in text, right? This is why music is a new is another language. And um, and the the cool thing about AI is that it can really like analyze the audio elements, right? And then trying to find um similarity between audio themselves. So there's something say unspoken similarity. But we can when you hear it, you know that they're similar, but you can't you can't articulate so much into metadata that the how it's similar, right? And uh, but the thing is like I can can learn this pattern and then uh cater to to that discovery experience.

SPEAKER_00 14:50

Wow, that's such an interesting term you just used there. Uh unspoken similarities. Like that's that's that's genius. Like that's just you've just hit across something there that's like, yeah, that's a major problem, like every day for loads and loads of people. In in in you know, in studios, in in in in trailer houses, in all sorts of different places, you know. That's that's that's an issue. Um, so would the goal therefore be to license that software or to to make that software available on a sort of B2B level um for those companies?

SPEAKER_02 15:29

Yes, yes, this is what we are trying to do. We we build our own platform and also we have the API so that for companies that are more tech native, they could use the API to integrate this technology into their own software or platforms, and then for those that are say less tech native, right? Uh, we build the platform so that they can just upload their music on our stuff and uh our platform, and then we will host it and we help to you you can share your pages, say, and then uh share your playlist and all this, and then people can search, but really using this very intelligent um search um yeah, like a function of like uh yeah, smart search system, basically.

SPEAKER_00 16:23

That's amazing. That's oh my god. I'm just thinking of the like the amount of music supervisors that I've spoken to, even just on this podcast and outside of this as well, where where they will literally like how they're building music, they're looking at so many like temp tracks and stuff that they can't license that they want to but can't afford. Um, and then it's like, right, let's get something similar to that. And so that might then be sent off to like let's say an agent to go right, or a publisher and go, right, can your writers write something like this and then they wait on it for another week, and then by then it's like, oh yeah, no, it's not quite what we meant because it's been reinterpreted or whatever. Which is in some ways is great because that's where the art comes from. But when it comes to media production companies, it's like that's not useful and they run out of patience, and then they just go, Oh, let's go over here and license this production music, hence why production music is in such a boom right now. But like you mentioned, you mentioned Fred earlier, Frederick Schindler. Um, and like I may I know what he's trying to do with with with all of the labels that you know that he's working with. Uh, and he's like, This is really, really underserved because there's like all this like amazing back catalogue from like some real great legacy indie labels, Matador, 4AD, all those kind of labels. Wow, it's like there's music that's like kind of untapped there.

SPEAKER_02 17:37

Yeah, yeah, yeah, totally. I I I think they have a great taste in their music creation and honestly super excited about it.

AI As A New Instrument

SPEAKER_00 17:46

Yeah, oh yeah, totally. Okay, so that's that's one side of it that and and by the way, I should say this, Tyengu. So if if you were only building that and just that, um like you'd do really well anyway, right? Okay, I think you'll do really well, right? So like even if it was just that, like that's like congratulations, like pat on the back for that, right? Um but thank you. There's another element to this as well, which is the the AI tools for music creators, that kind of extra instrument thing that you were talking about there. I um in a in a recent episode, I was interviewing um a gentleman by the name of Pete Ganbarg, who was president of AR Atlantic for several years, incredibly impressive guy. And I asked him um his his thoughts on like the subject of AI music creation, and he was like he had a very interesting take. He was like, you know, um it it's not a problem, right? It it's just another way of making music, um, and and in the same way that other things have been, like synthesizers or samplers or drum machines, whatever it might be. And I really like that take because it makes us realise that actually, you know, it makes everything a lot more inclusive, and it also means people can tap into creativity that they don't quite know how to articulate or don't know how to reach to or express. Um, so so that then leads us into the second part of what you're what you're building. Could you tell us a little bit about that?

SPEAKER_02 19:17

Yeah, sure. Um actually it's super interesting that uh that I have noticed this um too a very polarized opinion about like AI for music creativity that like say uh many of musician friends of mine like they they absolutely hate it, right? They they they will not, they will not even try. And uh and they they um and I totally understand it uh also like um because because how the tool, how the AI music tool today uh is working, and and this is what we want to really change. And uh because like you said, like um it can also be used as a sampler, as a synthesizer, as uh as a very interesting sound making machine. And all we do in music creation is to really find the sound to express, and then if we can have a much more versatile kind of instrument, uh how interesting can it be to for our creativity, right? Because um you you could be super, super um skilled in certain instruments, like uh oh, I play guitar very well and also piano, but your sound, the language of the sound is constrained to the two instruments that you know, and um and in a way AI can can break this barrier. Uh, I'm not saying that AI would just play guitar like that well. I I don't think this is actually the the path to go, but more like the path that um how can you translate this weird other sounds that you can't arc articulate with all these tools that you have, like the um the samples, the doors, whatever, uh the synthesizer. In the end, it's quite difficult to to really get there to to hear something that's exactly what you hear in your mind.

SPEAKER_00 21:30

Yeah, yeah, yeah. That that that makes a lot of sense. Okay, something happened yesterday to me. I was in a room where um somebody, it was one of my students, right, where the university where I lecture at part time, and um a student had played me a piece of work and it was um it was all midi, right? Um and like she'd created this outstanding piece of music, I thought. Um, and it was all like guitars and synths and you know, and and um and I said, uh have you thought about Swapping the samples in there, right? So because like not changing any of the MIDI data that you've used to create that, but just switching the entire thing into a sort of 1960, mid-60s style orchestra. And uh and she was like No, I I didn't think about that. I was like, okay, but but don't be offended, I said, because like we're not changing anything you've done, but just the sounds that you selected to drive what you've created. And she and so we tried it. And it was brilliant. And it's not, hey, look at me, what a great idea. It was more like being open to going, like, what you've done works, but how it's being presented could shift to find a different home, right? Um that's that's not quite AI, but it AI can do that without the need for some-yeah, yeah, right. Without without the need for someone like me going, Have you thought about doing this? You know. Um, so that's a that that could be an example where that is less offensive perhaps to music creators that worry about being replaced, for example.

SPEAKER_02 23:04

I I think it it shouldn't and will not replace music creators. Uh this is really my take. Because if you if you listen to all this generated music, like they might sound pretty real, like they sound like music, and um, I guess some people like it, but the more you listen to them, they all sound like the same song.

SPEAKER_01 23:27

Yeah.

SPEAKER_02 23:28

And uh I I don't I I think um the connection, me as a huge music fan, like um the deep connection uh that I I perceive from music is the soul of it, and the soul is in the storytelling, is in the creativity, is in also some the things that's um very personal, you know, like with a character. And and this thing, I I don't think um currently we we could generate songs like that. And um and I think like if uh we use AI as um creativity tool, we need to put human in the very center of it because the ideas, like no matter what you put into it, the idea is from you, from your experience, from your story. And this is something that's that will never be replaced, I think. And the taste in it, because you can generate things, but you would select what would be accepted. Uh that that that yeah, like you you can say I have this idea, and then AI generate this, and then you hear it, and this is nothing, like what's in my mind. So you you wouldn't you will not publish it, right?

SPEAKER_01 24:44

And um, yeah, yeah.

SPEAKER_02 24:48

So this this is really how I see like I I think the um the technology is is really fascinating, but like before using this technology to design any tools, the intention of the tool is as important, I think. Um like are we building a tool that's trying to hack hack the system? Just just just for the you see.

SPEAKER_00 25:27

Yeah, yeah, no, that makes that makes a lot of sense. Well, one one of the one of the descriptions that you used was um I think it was I think I read this on on your website, uh, was about or uh creativity uh that can be reimagined. Um Yeah. Is there I I know it's still early, right? But it it are there any insights that you can give us in regards to what the tools will look like on Songpot for music producers and creators.

Humming Ideas Into New Styles

SPEAKER_02 25:57

Yeah, of course. Um we're still exploring.

SPEAKER_00 26:02

Okay, fair enough. Yeah, yeah.

SPEAKER_02 26:04

So I I I I can't already tell you like specifically, but say um so so one thing I have built that that's pretty interesting, which is um you can you can use um chord progressions or melody together with the description of what you want to have, and then it will generate exactly like with the same melody and chords, and and then try to imagine how it will sound like. So for example, you can hum, you can hum something, and then you say, I want this in um um classical romantic period or something, something that I cannot do, right? But I have this in my mind. I have this melody that I dreamt of, and uh I I want to hear how it would sound like, and then it can generate that.

SPEAKER_00 27:05

But oh, okay, that's interesting. Okay, do you know what that's just made me think of? Um, I was watching uh a documentary recently, um, I think it was on YouTube, and it was like it was Michael Jackson in the studio working with Quincy Jones uh on I think on the yeah, the thriller. Oh my god, it's fascinating, isn't it? And it's just made it's just made me think of what you just said because like you know when he like he had that like inane ability to be able to go dn all I I I won't pretend to do it right, I can't do it like he could do it, but um You're pretty close, you're pretty close. Oh well, that's very kind, and uh and then like you know, then Quincy Jones would be like reinterpreting what again, oh right, yeah, that that bit's like chords, and like that that that's the intro for Billy Jean, and that bit there that you just sung, that's actually like that could be a bass line, right? So let's turn that into a bass line. And so all that that that is that kind of similar to what you're referring to there, yeah.

SPEAKER_02 28:06

Yeah, yeah, yeah. This is the goal to have that Quincy Jones.

SPEAKER_00 28:11

Yeah, yeah. Wow. For you and me in the studio, yeah, yeah, because um, you know, if if if Michael Jackson hadn't have met Quincy, right, then Thriller wouldn't have sounded the way it did. And but not not everybody, you know, can can afford access to a producer like that, right? And um and so this is like w when people talk about like an artist-centered ecosystem, I think what that actually means is artists having the same tools as the business, um, you know, like having tools to the same marketing tools, the same playlist pitch tools, and the same tools to be able to go, okay. I I I can't, I haven't got the budget for that producer over there, but what I really need is to get this idea out of my head. Um, and if I hum it over here and then do this and that, and then I can get what's in here on there to the world, right? And that that's that's the goal, right?

SPEAKER_02 29:05

Yeah, yeah, but in the meantime, I don't think this tool will replace Quincy Jones because Quincy Jones has this mastermind that that maybe the machine can never can never achieve, right? And um even like I I I would say it's a better say prototyping way, right? Uh better way of prototyping. Like, say you you can have a lot of ideas and then you can hear them instantly instead of like getting mad at um oh, I'm trying to find this sample that should sound like this, and then I spend hours browsing the sample library instead of that. You you could just you see, uh to make drafts, let's say, but I think uh a really skilled great producer, I mean they they do master work. Uh yeah, yeah.

SPEAKER_00 29:55

I'm I I I'm so glad you just said that because um uh yeah, this is not to replace Quincy Jones, it's to be able to help the the next Michael Jackson to formulate those ideas before they take it to the next Quincy.

SPEAKER_02 30:12

Yeah, and uh actually I have been talking constantly with musicians and producers, and um after a lot of chats, I realized like one of the biggest problems in their production process is actually the communication. It's it's about how like you have something in your mind, you communicate to your producer, your producer don't get you, and and and and and you go back and forth, back and forth, but um, what if you you could quickly prototype something that sounds similar, but not quite there, right? And then you just send this as a reference, and then you could communicate much easier, also.

Tiangu’s Path Through Music AI

SPEAKER_00 30:52

Oh wow, oh wow, that's okay, that's amazing. Do you know when people often say that like music is you know the universal language that can travel all over the world, and like that's beautiful, it's amazing, but it's also a very complex language, isn't it? That can be misinterpreted even between people that speak the same language. Uh so yeah, I I I I love the I love the sound of this. Um let let's let's shine more of a light on on you and your background, right? Because you have a correct me if I've got this wrong, right? You got a PhD in music information retrieval, um, and you've you know you've won multiple awards as like a very competitive programmer, and and you know, you contribute towards you know various pieces of literature and books. Does your does your your passion for sort of science does it play a big part in your curiosity of how all this sits together?

SPEAKER_02 31:47

Uh yeah, definitely. Uh not just limited to my curiosity for science, I guess, just curious curiosity for for this whole world. Um yeah, yeah, and uh actually, yeah, I I am definitely a super geek girl, let's say. I I started programming when I was seven years old. Really?

SPEAKER_00 32:12

Yes, seven years old.

SPEAKER_02 32:14

Seven years old, yes. Wow, and um yeah, and and for a very long time I was the only person who knows how to program, let's say C in my whole school. And wow, you were you were that kid. I was that uh weird, geeky kid. Yeah, but in the meantime, I I mean I I love music. I I just I don't know, my my whole uh it is the biggest passion of mine, and um yeah, I I remember like the first time I I've been to a music festival, it was I was 13 years old, and I was so fascinated by how it feels, how how how everyone just connects with each other, like gather together for the music, like really how music connects us, right? And uh this experience, but also I feel understood a lot of times by just listening to the music I love, you see, because they are expressing something I want to express in this beautiful language of music, but I I couldn't do that, but they did that and and I love it.

SPEAKER_00 33:27

Um what a wonderful, what a wonderful way of explaining it. That's I love I love that. I love that. That's that's terrific. Um I I saw on your um on your LinkedIn page that you recently attended uh ADE uh in Amsterdam. Um, how was that? What did you learn about the EDM sector whilst you were there?

SPEAKER_02 33:48

Um in 80 actually, I I didn't go to that many events because I I went to mainly the panels uh to listen to people in the industry, how they see everything's going, and uh you know, but um definitely the electronic music scene was huge in Amsterdam. You just walk around and uh you pass by a flower shop, maybe, and they they're having a DJ they're playing. It's such an experience. Or a cafe shop, you know, like the whole city is celebrating electronic music, it was such an energy. Yeah, and uh also I I met so many like talented musicians and producers there, and we exchanged on what kind of tools they would like to have, right? And and they definitely gave me some very interesting feedback for songpart and and also shaped kind of the product that we are going to uh ship.

SPEAKER_00 34:51

Oh, very good. So it kind it was kind of like a research mission then research mission, yeah. Okay, okay, okay, so that's very cool. Because you're you're still you're still in research and development, right? For the for the creator tools. So you're yeah. Okay, so you're kind of going, what do people really need? What do they have problems with? Uh, because you're good at like fixing those problems, right? So once you know what the problems are, then you can go, ah, okay, I know how to do that.

SPEAKER_02 35:22

Yeah. Um for the last seven years, uh, well, like before I started SoundPod, I was really just doing music AI research. And for many years it's a super niche uh topic. And actually, the musician around me, they they don't know about what is music information retrieval, and uh they didn't know that people are composing with AI, that this was such a niche uh thing for many years. And I have experienced, like say, five years ago, I remember I I listened to a talk by a researcher working on generative music project, but when he plays what this machine generates, I didn't thought that this would go so far, you know, compared to what we can hear today. And um yeah, so uh after spending so many years like really digging into the tech that's that's behind uh this music AI, I want to build something that's actually useful for the industry, you know, not that that um with my understanding of this technology because I think I I yeah there are so many cool things to do there.

A Ten-Year Vision For Translation

SPEAKER_00 36:42

Yeah, yeah, that that that's nice. Um wow Gengyu, you are you are fascinating. You really are. Like I can see exactly how useful you're going to become in the coming years to so many people in so many different sectors, actually, as well, of the music industries. Um, you know, what what are your what are your goals moving forward? Like, what do you want this all to look like, you know, sort of, I don't know, let's say 10 years from now? 10 years. It's a long time, innit? But it goes fast, and it's a long time. Time goes faster as you get older, I find. So I mean 10 years will fly by, but like uh right now it feels like a long time, doesn't it? But like, what do you want to see going forward? Do you want to just it's the goal to find solutions to people's problems?

SPEAKER_02 37:33

I hope in 10 years we can actually solve this problem that's that I described in the very beginning of this podcast, right? Like to how to translate that sound in your mind to the sound that you can hear. Um, I I hope that some part would play a big part in solving this problem. And I hope that we could really build something that's um that's serving the creative community. We we we we help the creativity uh with AI, right? I I really hope um yeah, this this is the wish and the goal that I'm heading.

SPEAKER_00 38:14

Yeah. Do you know what I I I'll share this thought with you because before I before I spoke to you today, having had a look at your work and and then hoping like again, thinking, oh, I hope she can come on the podcast. I'd really like to talk to her. So I'm really glad that you that you have, right? Because one of the things that I was thinking when I first understood your work, I was thinking, right, so you're looking at AI in regards to like optimizing it for catalogue and discovery, and then AI in terms of like how it can help to create music. And I was thinking, I thought to myself, that's wide, like those are like two different areas. Actually, in talking to you, tell me if I've got this wrong. I don't think they are. I think what you're not no, they're not. You you you're creating something that can do the thing that people can't explain, right?

SPEAKER_02 39:06

Yeah, I mean, uh, in the end, uh, if you build an AI that understands the language of music, wow, it can do both.

SPEAKER_01 39:15

Yeah.

SPEAKER_02 39:16

Yeah. Uh this is how I see it. Like uh, it's two sides of the coin. But um, yeah, I I understand, I totally understand that it looks like um two different uh products or even two different companies working for two different sectors in the music industry, but from the standpoint of the technology, no, uh if you build a AI that's powerful enough for music understanding, you you can solve both.

SPEAKER_00 39:48

Yeah, yeah, I I I I totally agree. Well, I'm I'm convinced. I'm so glad that like that you you you were cool with like coming on here today and talking about about this because uh I'm so excited for you, honestly. Like, congratulations on where you're at.

SPEAKER_02 40:05

Thank you so much. It's very kind of you, really.

SPEAKER_00 40:08

Oh no, no. I it's I'll tell you what's kind is what you're doing, right? Because to be quite honest, right, you are an utter brain box, right? You've got so much ability, so much cognitive ability, you're such a blessing to this world, Tianka. You really are. And you've decided to plow all of that into making the music industry better. So when it comes to kindness, that's kindness, and thank you for what you're doing.

SPEAKER_02 40:36

Thank you. I I really hope I can make something useful for this industry. I yeah, uh, I would love to contribute a bit.

SPEAKER_00 40:44

I I think I think you already are, and I think that you'll do it more and more as time goes by. And I'm really excited to see how it all unfolds and and to see what happens next, and to see what problems you solve and and what happiness you bring to people uh in their work, in their lives, and the impact that has on music consumers and fans, right? Because we're all music fans, that's how we ended up here, right?

SPEAKER_02 41:10

Exactly.

SPEAKER_00 41:12

That's what brought us all together, isn't it? You know, um I'm I'm I'm fascinated by by you and your work, and I I just wish you just all the luck and all the love going forward, and thank you for contributing your time here today.

SPEAKER_02 41:25

Thank you so much, Johnny.

SPEAKER_00 41:28

Okay, wow, what a fascinating lady. Um absolutely fascinating. Uh, she clearly has a solid understanding as to what it is that she's trying to do, and like many good businesses, businesses generally look to try and solve problems, don't they? And I think if you take Tchengu's love for music and her ability to problem solve, what she's created is something that's absolutely fascinatingly useful for quite a wide range of different people. I'd be very interested to see how her product range actually evolves as well, because right now it might look like, and I said this to the here in in in the interview to her that you know, does it is she building two different things here? And actually, that was my concern before I spoke to her, and I am no longer hold that concern because actually what she's building is the same thing, it's the same problem, the ability to be able to understand and explain music, it's always been a problem. I can tell you, as a producer, sometimes there's an AR and there's an artist, and one's saying this and one's saying that, and then there's me as the producer in the middle, kind of going, I think they want this, or I think that he wants that, but like I don't really know, and we won't really know until we get there, until we've made it. And that is the art, and that's fun, but it's also problematic. I think what Tiengu is building kind of looks to try and problem solve things like that by be able to take a tool and take some of that subjectivity out of it, but leave the art in it. Now, that is incredibly complex to know how to do that um and to how to communicate those things, but I think that she's doing it. I I think she's doing it, and I think big things are ahead. So I hope you enjoyed listening to that episode. I I I I hope you're as fascinated as I am about what happens next for Songpot because I'm very excited about what that could uh how much time it's going to save for a lot of people, uh, and how it doesn't step on the toes, on the hearts or the minds of creators, right? Uh, if AI does that, it can be very offensive, and I'm not here to advocate that one bit, and neither is Tiengu, right? So it's about looking past that and looking at it and go, actually, this doesn't do that. It's this assists people, this saves time, and actually it achieves the goal of making music and making music discoverable and accessible. Okay, that's enough from me today, but I would like to revisit Songpot in the future just to keep an eye on what's going on, because I see big Things ahead. Anyway, that's enough from me today. Till next time, everybody, have a great day, and may the force be with you.