#149 Alex Klein: The Next Wave of AI To Rock The Music Industry

0:00The time now is very much like at the beginning of the internet where there are organizations that are just going completely AI native that live and breathe AI that are thinking about how to bring it into all of their processes. and then there are organizations that are like, oh yeah, we're aware of this and we think we may have a strategy around it. We want to make products that we can use ourselves and really enjoy and have no qualms with using them above any other competitor products in our potential price range or lives, products that we can recommend to our friends and family. Hi, this episode is sponsored by Salonis, the global leader in process mining.

0:38AI has landed and enterprises are adapting, giving customers slick experiences and the technology to deliver. The road feels long, but you're closer than you think. You see, your business processes run through many systems, creating data at every step. Salonis reconstructs this data to generate process intelligence, a common business language. With process intelligence, AI knows how your business flows across every department, every system, and every process. With AI solutions powered by Solonis, enterprises get faster, more accurate insights, a new level of automation, and a step change in productivity, performance, and customer satisfaction.

1:26Process intelligence is the missing piece in the AI-enabled tech stack. Search Celonis, C-E-L-O-N-I-S, to find out more.

2:01other digital content exposed, demand superior content protection, and maintain trust in your brand with Digimark. Hi, I'm Craig Smith, and this is Eye on AI. In this episode, I talk to Alex Klein, founder of Kano, a company that produces the Stem Player, a smart speaker device that uses AI to separate song stems and allow new ways of listening and remixing music. Klein recounts how he started with educational do-it-yourself computer products before meeting Kanye West at CES and collaborating with him on the stem player. The conversation provides an insightful look into the intersection of music, tech, and AI.

2:50I hope you find it as fascinating as I did. So I'm Alex Klein. I'm 33. I live in London. I've got a team here of 30 people. We've been working together for over 10 years now, many of us. and our company creates products that are designed to leverage the most advanced technologies to relieve people's suffering on a day-to-day basis and how the idea of stem came about we actually did not come to music via targeting music as an industry we were the very first product we made um back in 2013 was a bright orange box it's actually behind me above my head filled with color-coded computer components and a storybook a page-by-page storybook that would show you how to put the components together um it was a stem computer kit it was a stem toy it was one of actually the first stem products that would teach kids about stem so you could buy it from toys r us buy it from target uh it was originally the first user was my six year old cousin mika who wanted to build his own computer um and those sold under the brand name kano kano was named after the founder of judo who was a japanese primary school teacher named Kano Jigoro.

4:34So that's like chapter one for us. We have this notion that, you know, we live in this world. I used to be a journalist, actually. So I used to work at The New Republic, Newsweek, Daily Beast, written stories for New York Magazine, The Nation. I was covering things like um occupy wall street um um mitt romney's uh tax returns the church of scientology just a wide range of sort of political economic some tech instagram's acquisition have had a sort of passing relationship with the times actually craig because uh uh jill abramson was one of my journalism teachers in university oh yeah really for real yeah where was she teaching she she taught a course at Yale.

5:22She taught a German course at Yale. This is while she was editor or before? This was, I believe, before where she at the time was deputy editor or she had... Right. Yeah. She was deputy for a while. Yeah, exactly. Very interesting person. So we went on a tour of the times and I've always had a... In fact, I applied for an internship at the times in university was rejected it's unbelievable you know my my career might have gone very differently uh had i been accepted but um yeah yeah you're probably um it's probably good that you were rejected rather than getting sucked into uh into the vortex of the times yeah the the journalism world was super um welcoming to me at that time i was like a college student and i was able to get a few internships here and there.

6:21And I was into computer science and I was using early data collection techniques to build infographics and that kind of stuff. But I think the kind of spirit of like being a working journalist, I was at Tina Brown's Newsweek Daily Beast kind of carried into STEM, you know, which was the idea that, well, what if you could build a technology technology product and a technology ecosystem where the idea wasn't just like you get delivered of this and you just it zombifies you what if you could get a piece of technology that you could actually see how it works and put the pieces together yourself and participate more in the technology that you use every day and so that idea carried through to what eventually became the stem player which uses generative AI to generate the pieces the puzzle pieces of any song you throw at it and then you can just isolate the vocal isolate the drum or even put the pieces of different songs together as simply as putting together legos or even automatically so uh yeah that's the journey from kind of journalism to stem computer kits for kids sold over a million of those around the world worked with warner brothers disney microsoft was an investor and then three years ago did our first music product the stem player which eventually became a collaboration with uh kanye west um that's how a lot of people discovered the device yeah i want to back up a little bit the the uh the storybook uh stem uh construction product i mean you know i'm i spent most of my adult life in china and i i was very early in my career in the garment trade we're manufacturing in china so i have a familiarity with uh contracting manufacturers to build stuff and as a matter of fact these glasses which are red because I lose my reading glasses all the time.

8:33So I buy them red and I can always spot them around the room. And I got tired of paying$15 a month,$15 a pair online. And I went on Alibaba, found an eyeglass manufacturer, found one that had a style I liked, picked the color, and I bought a minimum order of 120 pairs. So I've got eyeglasses for the rest of my life. But my point is that I'm kind of familiar with that to me is not a huge barrier, having an idea and then finding a manufacturer to make it physical. And I've done that, you know, as I said, in the garment trade and different projects in my life. But, you know, that's not obvious to a lot of people.

9:26A lot of people see that as, you know, a lot of people have ideas, but actually manufacturing a product is beyond their imagination and capability. So how did you come about doing that? Yeah, it's a great point. I mean, at that time, so I come out of journalism. I'd gone to get a master's degree in Cambridge. So I was doing an MPhil in politics, maybe with a view to like going back and continuing writing in politics. and my cousin at the time was you know he'd been like a sort of tech forward internet guy for a while journalism as well he was like involved in the telegraph bringing the telegraph to the internet and he knew of this technology in um in cambridge called the raspberry pie um which you might have sure yeah so i'm he put me onto it and then i sat down with ebon upton who had effectively invented the concept of the Raspberry Pi, taking a development board for mobile phone demonstration, basically a reference board, and redesigning the layout so that you could, you know, basically run a full operating system with full Linux distribution on it.

10:40And it had started selling like hotcakes, largely, I think, because people were doing something similar to what you did with the eyeglasses, but with sensors and buttons, buying those from, you know, markets from Alibaba, from the East, assembling their own sort of maker projects at home with the Raspberry Pi as the brain. So doing things like, you know, making a, like a submarine, like an autonomous submarine, making like a weather balloon with a camera in it and taking pictures of the earth and running Bitcoin miners and different clusters. And it was this whole sort of cottage enthusiast homebrew computer club situation.

11:21And there was this guy at the center of it called Eben. He was at Cambridge. I was at Cambridge. So my cousin was like, you should go talk to him. And I actually went to interview him like for a story. Like I was going to pitch a story on this back to Newsweek. And I was just interested in it myself. and he shared with me all these challenges about he basically you know there's so much demand for this but people people have trouble understanding it you know people have trouble getting into it and sort of the success that they had had manufacturing this plus the challenge of accessibility that the product had made me think like you know I didn't really know what I was going to do next.

12:04I was like, well, what if my cousin was, was like into the idea and into working with me on it. And, you know, he had a bit of capital. And so he was like, well, why don't we do this? Like, I will invest in this thing and you build me a product that like me, you and my six-year-old son can build together. You know, you, you know, design, you know, tech, you'd like to write, write a book, like write a book to put the pieces of this computer together. And that was the first product. It was like really a family affair, like, you know, on the living room in their apartment. Met up with friends, showed it to them.

12:42I was introduced to this Israeli guy who had operations experience from a company called Keter, an injection molded plastic chair company in Israel. That kind of like symbiosis of our relationship worked out really nicely. and uh yeah we launched the first product on kickstarter which was a huge thing back then it broke all the yards and that kind of launched the mass production of the first stem computer devices so i think i was fortunate to meet the right people and to um uh have to kind of start my company at a time when people were starting to get into the idea of electronics manufacturing on a like hobbyist scale, which wasn't what was definitely not possible 20 years ago.

13:29And just went through a big jump in accessibility when things like the Raspberry Pi and other things came along. Yeah. And what were you doing at Cambridge? I studied political economy. So I was studying public debt, treasury, the US treasury, the UK central bank, and the the kind of relationship between high levels of securitized public debt in what were called at the time developed economies and growing levels of socioeconomic inequality in society and looking to see if I could gain enough of an econometric understanding of the two phenomena to make an argument that they were connected. So it was sort of deep economics research which around a sort of counterintuitive theme that like greater public debt equals greater inequality when uh you know typically it was thought at the time well you take on a big public debt to equalize society have a welfare state and system that you know gives everyone a safety net a fair starting point um and i was i was sort of arguing that in principle yes but in practice the large um sprees of um monetary base expansion that the central banks had been on had not been channeled into, you know, welfare improvement programs in the way that they had been described and pitched to the American and English people as, but were really more about inflating asset values.

15:09And I think, yeah, that was i was like very much like considering a more wonky trajectory of for my life for sure i still try to stay pretty wonky but there's definitely a lot more like yeah and was that a graduate program or were you undergraduate or what what what's the education that you've been through it was that was an mphil um so like a master's of philosophy which is like a one year at cambridge after you do your undergrad my undergrad i got a ba in ethics politics and economics um but i minored minored kind of in in in writing and journalism and computer science was always kind of a hobby thing for me i never really expected it would become my my career right yeah okay and then uh this is fascinating and then stem uh the product uh yeah describe again how you came about building that it was that directly from the storybook uh product yeah exactly so we this kind of product formula of like there's a a box and And inside there's like these beautifully like designed pieces and you put them together.

16:32Then the device like shows you how to put together some blocks on a screen and you make something that this construction-esque step, STEM kit slash software-esque things. We'd done them in all these different domains. So we'd like, we'd done a computer, which was like something you plugged into your TV. We'd done a tablet, which had a touch screen. We'd done a magic wand, like was literally an electronic wand that you would build and then cast a spell on a screen. We did a board of lights that you could like paint on. We'd done a camera that you could actually put the camera together and then change out the lenses and do digital photos and make your own Instagram filter.

17:14So it was all like making, like the slogan of the company was like, anyone can make, you know, anyone can make. You might think it's only for some, but anyone can make. So we then, we applied this idea to music, and we'd been experimenting with machine learning for the Harry Potter product, because the Harry Potter product effectively had to filter out a lot of motion data in order to capture the precise motions of the Harry Potter spells. So Wingardium Leviosa, etc. etc we went to do a music product next called the speaker kit or the stem speaker um that was like this kind of orb like thing you would put the top and the bottom together and then there were these knobs on where you could like control the stems as they the music as it was playing through you could kind of filter the music up and down and that was ultimately um the product that i think helped solidify um our relationship with kanye because when we saw it was when he saw that was up and running he was like can i put my album on this because he he always liked our design you know when i first met him at the consumer electronics show in las vegas um he'd like picked up our computer kit and he was like this is fresh um uh can you give me this dope clear tablet joint and we got on the phone and you know he had a real enthusiasm for for the look and feel of of what we were doing and um he said he liked the simplicity and he liked that anyone could understand it and um yeah i mean it was a very uh uh magical moment uh for me because i'd always been such a fan of his music and i'd seen him as a really you know inventive thinker and i you know But at that time, he'd had a lot of criticism because he'd endorsed President Trump.

19:13And that was always a concern for me because, you know, I feel that President Trump is a fascist. So when my favorite artist had endorsed him, that had always been a concern for me. um but i i kind of like put it aside for for years and we became friends and we became um kind of uh creative allies you know i would i would help him with his music i wrote lyrics on one of his albums strangely enough i'm a lyricist on jesus is king yeah which was so weird and so great and a blessing and he would help me think through some things you know um about my business about how to present the products, how to communicate the products.

20:05And that was fruitful. We worked on many different products, actually. Refrigerators, lights, appliances, speakers, headphones. We touched all sorts of different domains, even telecommunications. But what ultimately came out was the STEM player. And interestingly enough, because this is the Eye on AI podcast, the thing that in some ways contributed, at least from a product perspective, to our kind of exiting the partnership was not only the terrible things that he said and the terrible, hurtful, anti-Semitic things that he said, although as we say in Judaism, that would have been enough. But it was also, he was very opposed to the use of AI at all in the product.

20:58And I always saw it as a very magical element that you could bring any song, separate it into the stems, hear it in this new way. His view was it should be just Kanye music on my stem player. And that to me was kind of a place where we started to kind of move apart. I remember one quote recently that I heard was that the time now is very much like the time at the beginning of the internet where there are organizations that are just going completely uh ai native that live and breathe ai that are thinking about how to bring it into all of their processes and then there are organizations that are like oh yeah we're aware of this and we think uh we may have a strategy around it it's very similar at the beginning of the internet and i think we were just you know we'd like to believe we're in the right camp and thinking that something truly transformative has occurred um in the certainly as you put it the the applications of the fundamental research and you know three years really three years ago the major breakthroughs were made and now they're they're starting to be instantiated in the products yeah let me just ask a couple of questions before moving on with where you're going just for listeners the the the product on the shelf behind you.

22:20The STEM in that product is the acronym for science, technology, engineering, and mathematics. But the STEM in the STEM player refers to, is a musical term, right? For, I don't know, is it one instrument or one group of instruments or sounds within a larger position. Yeah, maybe I can even share my screen here. Is that within the rules of the podcast? Yeah, sure, yeah, go for it. But, you know, the stems of songs are the pieces of the songs. They're the ingredients, you know? So the vocal, the drum, the bass, the instrumental, you know those are the those are the stems and so our technology allows you to split any song into those pieces and hear them independently so you uh you get the basic idea which is if you go to stem.com you can hear the vocals the drums the bass the instrumental of any song either by itself um or in combination with the others in any way you choose so it makes the music uh a deeper listening experience um and also we can get to this maybe later it also allows for songs to be combined in new ways yeah and uh and is that using uh some machine learning algorithm to separate the channels that it's uh i've talked to people on birdsong for example i did an episode on isolating individual bird songs within a cacophony of the forest yeah so the yeah the the the problem of um what's called in in audio uh like source separation is is basically the same in in those two instances you're looking to you know audio comes into your ears and you know there's just you know millions billions depending on what scale you're looking at of different frequencies that are all intersecting and but certain um human recognizable things have do cluster around certain frequencies so for example you know voices tend to be of a higher frequency than um percussive elements or bass elements in a song um you know the bird songs of a particular species of birds may cluster around a similar recognizable set of frequencies so um source separation has actually been a thing for like 50 years even since like radio is around like i'm getting this radio stream in can i just in an analog way um filter out the the signals and get kind of different bands of the audio and either even like security and radar based, either security and radar applications of some of the same technologies.

25:38The big change came really in the last three years to source separation when generative machine learning techniques were applied, effectively supplementing the weaknesses of traditional signal-based source separation, which is to your ear, a machine just from an analog perspective filtering out a particular band of frequencies, even if that band dynamically changes with a bunch of subsequent parameters, your ear will categorize as a voice just ranges of frequencies that are more nuanced and subtle than can be picked up by just a brute forced sorting algorithm. So the generative approach is basically to start with source separation and then use tagged data, so data that's been tagged as vocals, drums, bass, or instrumental, or I guess as a finch, a skylark, or to supplement and actually fill in what's missing out of those analog separated stems and techniques even from the synthesis of voice like google's wave net architecture which is now used to in most voice simulating technologies can be used funnily enough to fill in the missing parts of like a drum or a base because there are some similar qualities, it seems, in how humans perceive the different parts of music to how they perceive the human voice.

27:17So those were two of the big breakthroughs, using kind of a more generative approach to traditional analog separation, using techniques from human speech synthesis in the prediction objective functions of that technique. And on top of that, there are also improvements that have been made in just data handling and data processing that make it faster to actually get those stems out of the model and to the client, whether it's like a web browser, an application or a piece of hardware. So that sort of stem separation is at least as we do it and as it exists today. and uh i only have a half an hour left i have a hard stop at five there's so much to talk about so first of all you're at ces showing a bunch of products yes you're at a presumably at a booth and you know you got your signage up you got products and people are coming and what kanye west just walks up or is it was there like one of his people comes up says hey this might be interesting would you like to meet with kanye so one thing i learned about yay over the years which was really uh i'll answer but it's like he does a lot of stuff directly himself in a way that is very endearing um because you know we all know this world famous sky but he's very much like he's raw he goes at it himself which is so the the actual like i'll set the scene um which is i was at ces but i was actually not at the booth on the on the day kanye came down so i was in my hotel room i was in my boxer shorts i was kind of sleepy and i was catching up on emails and you know thinking okay it's the last day no one's coming through really we've both gone through the big days and my um my colleague and friend Vidal at the time just text me I pick up my phone and it's a picture and it's uh the table in front of our booth we had a nice big area we'd done the thing with Warner Brothers that year so a nice big area two two walls table with all the computers and at the back of the table there was just Kanye standing holding looking at the computer and I was like I was like oh damn I was upset because I was like Kanye is my favorite.

29:53I'd been interviewed in 2015. And I'd said, if there was one person I could pitch my business to who I haven't pitched yet, it would be Kanye. And I was kind of annoyed, but also look at that. I wasn't there. I was like, oh, but then I was like, okay, let's see what we can do about this. So I called my other colleague, Greg, who was there, who actually studied at Berkeley College of Music and is a great drummer in his own right. And I was like, Greg, what's going on he's like yeah kind he's just here and he's you know he says you know he loves the tablet you know he said get me that dope clear tablet joint which i'll never forget and i'm like yeah we'll get him a tablet and i was like yeah we're getting him one like i was like hey you know go put him on the phone and um he was like i don't know how it's like there's people gathering because he'd now spent a long time at our stand people gathering people realized who he was he only had two sort of handlers with him and um greg was like i don't know i can't i don't know if i can get to him and i was like greg just walk through the crowd walk past his bodyguard and hand him the phone and say alex the founder of kano wants to speak to you and uh he's like all right i just hear this rustling and um you know i i hear this like kind of voices in the background and i hear greg's voice and then i hear more rustling and i hear like hello and i was like hey is this kanye and he was like he was like yeah i was like oh wow nice to meet you and he was like yeah man nice to meet you too i love what you're doing it's fresh i was like oh well coming from you that means so much what do you like about it and he's like i love the simplicity i love the transparency i like that anyone can understand it i asked him a couple questions about his kids how they deal with technology and he was like it was very loud we couldn't really hear each other and then it was like hey i'd like you to come to calabasas to work with me on education farming the next generation of technology we're building a team and this was very yeah you know he was he was always like doing all the he was always thinking about all these different things and um you know I went to go meet him um at his house for breakfast and the meeting got changed locations and you know Chris Jenner stumbled into the meeting and so Kim was there and the meeting got redirected and you know yeah and i just kind of like he was looking at all our pro because we always make tons of products i've worked with the same people for a really long time my colleagues like bruno james tom we've just been going for it and even though our first thing was for kids the whole idea was really always about more than just kids it was about really stem as perceived as most broadly it was about the idea of like you know the world being something you can take into pieces and put back together again.

32:42And I think the project with Kanye gave us a lot of confidence that the quality of our product design and the quality of our team and business was such that the biggest names in the world were interested in getting involved. And then the advent of a solution to the problem of music decomposition around the same time, plus with this Kanye relationship just kind of came together to make stem player something that um you know obviously has set the stage for many other innovations in the space and i think now has given us a new platform to grow from as well um great okay and uh i mean i i have more questions about uh the kanye west episode but so you build this product he releases his album on it the product blows up within a certain uh market uh did it go through an arc is it still out there and then i'd like to hear what you're what you're working on now absolutely so yeah no the product um it was a really interesting story because there were so many different projects that we put together with yay and of course as a business person i was like i wanted to finish one of these projects with yay like i because we had a good relationship you know we were able to get a lot done but he he always struggled with finishing stuff and i felt like he would bring me into the equation when he wanted something to get finished and get done because i'm you know i'm the bring me in and like basically it was around this album donda you know we went back and forth you know we got in arguments i went out to wyoming i was on the ranch with him driving around in atvs and um the stem player at the time wasn't even our main focus we were focused on a different device the stem player you know we started a project with him but then the project got cancelled and the contract got terminated and it was like we were focused on something else so we ultimately decided we weren't sure if we were going to be able to move forward with kanye at all just because he wasn't really like um first of all from a business perspective there were there were issues and then it just didn't seem like although i really thought he was creatively gifted that he was actually going to get anything done so i proceeded to just take out the stem player as you know our own product and it was during that time that we really emphasized and developed the the ai element that it was super against because we were like well if we're not going to get a kanye album on this we're going to just create it so that anyone can easily and that's really the direction that we've taken since right but but kanye um did ultimately reach back out around donda did ask about let's put something out we had devices ready we designed the exterior of the device to kind of work well with what he was doing at the time and then he right before the release of donda put um his album on it that was non-exclusive we sold um a good number of units just as pre-orders because people were excited it's the first time they'd seen the device kanya gave it a big stamp of approval from an audio and music perspective and fashion culture perspective then during the rollout of his second album of that cycle donda too he actually got to a point where he was so frustrated with his relationships with the record labels and with um with the major streaming platforms that he decided he would put uh his album only on the stem player um so it was completely exclusive we worked on all the communications and the content and the marketing.

36:44Kanye, you know, put it on our device because I think he felt it was a better way to experience his music. You could hear the music in this new way. And I'm very grateful in a certain sense to him for, you know, doing that. But I think also on his part, he was smart. He saw that this was like a future thinking thing. You know, the device ended up on like was everywhere. It was sharing about it like um people were using the stem player box to propose marriage to their fiance putting the ring in the box it was like it was a great it was a great time um and that that led to a certain wave um we ultimately had to dissolve the partnership but the stem player continued to sell uh we sold out of the first uh hundred thousand that we made um and uh we've been preparing to release the stem 2 uh this december okay uh now one question uh why the hardware i mean this could all be on a on a web interface well the the i think with the stem one the stem player there was a collectible quality there was a sort of um i'm holding the music in my hand And I'm using this new kind of interface to interact with music in a new way.

38:11It was almost like an art piece. You know, it was like it was very it was a very conceptual product in a way. It was like you get this and you get the album and you can listen to the album in this new way. Like you can flip it, reverse it, chop it, loop it. But I think for the direction of STEM for all, for accessibility, we knew that it would soon become a technology and a service that you could access on other people's hardware, as we like to call it, your MacBooks, your iPhones. And we did go in that direction. And we do have a service where you can enjoy some of the magic of STEM audio now in your browser, on your phone, wherever you are.

38:54and that's under frequent iteration and development with a good number of active users. The reason for the hardware, though, because we haven't abandoned hardware, is we do see some opportunity, frankly, in music audio hardware today. We see an opportunity to create devices that feel different to the existing devices that are softer and squishier and more like an extension of your body, you know, that are like one piece and can stretch and bend and are malleable, more like, again, an organic system rather than a technological one. And inside them, a more organic style intelligence based on a neural network, one that can take what you love to listen to from the Bluetooth stream or even directly on the device and augmented it and transform it just subtly.

39:48So you're still listening to the, you know, to the Beatles or the Beach Boys or Playboy Cardi or Little Dirk, but you're being introduced to new music because it's being mixed in. STEM is actually being mixed in. And we see that as a new form of music discovery enabled by AI and actually a new way for music to be monetized. So that's probably the last reason. From a monetization perspective, the whole music industry is like pretty janky right now. Stuff is not working. You know, the pie has not expanded enough and everyone's just competing for space on these tiny little squares on this tiny little bigger square.

40:27We think if you can make a speaker and a headphones connected to the internet where you can experience music in a really unique transformative way, artists are going to want to distribute music to those devices as well as these ones that everyone else has. And consumers may be willing to pay a premium to hear music in this new format, which brings you closer to the music on these new devices. Do you have one of the STEM players within reach just so that I've seen them, but I'm not sure everybody listening has. Let's open it up here.

41:08And then on it.

41:27Speed up. Slow down. Loop.

41:38I'm the astray.

41:43Stretch my hands to you. Life like this is what your life like. Try to live your life right. People really know you push your buttons like type right. This is like a movie but it's really very lifelike. Every single night right. Every single fight right. I was looking at the grandma and I don't even like likes. I was screaming at my daddy told me it ain't Christ like. I was screaming at the referee just like Mike. Looking for a bright lights. Looking for a bright lights.

42:13Yeah, that's fascinating. And the new one, so you're working on a new iteration? Yes. So without revealing too much about the new one, so with this device, the music comes loaded on it, and if you want to add new music to it, you have to pick the cable, plug it in, and plug it into a computer and then you can access that web interface where you can manage the content on board add your own music to it but we really want with the next device to remove that umbilical cord to the existing ecosystem so the stem 2 will be slightly larger but it will also be a bluetooth speaker with a much louder just gorgeous sound i think better sound from our ears perspective and also we had some amazing artists work on it the nicest fullest sound profile in a very unique form factor that you can put your existing music onto via Bluetooth they just play music through it like any Bluetooth speaker and it will transform that music into stems live and in real time so wow yeah that's you know maybe I did just give away a lot actually but that's you know you heard it here first um that's stem two's uh stem two's power really should be um something that takes your everyday listening experience and elevates it rather than being something you know kind of special and precious and that has maybe a beautiful applicability when you take it out of the box right wow uh and and so in in is the ai in it uh are you building because so much is happening uh certainly since this uh first iteration uh are are you using any of that uh new technology the the the uh sort of the conversational interface for example or or other forms of generative ai yeah i mean there's a there's many forms of of ai but um that are kind of included or in the product you know depending on how you define ai i suppose but um for the new product at least in the way in which it's advanced is um the stem separation uses a hybrid transformers model um not unlike the transformers model that chat gpt4 and 3.5 are based on um the um The generation of mixes uses an AI beat detection model that allows you to find the downbeat throughout a song, even if the beats per minute change.

45:20We also use machine learning, I'd say deep learning techniques to filter through your library to find songs that will match together and will mash up and sound good together. And then the data set that the system is trained on has grown because we've generated so many stems since then. And we haven't performed a new training yet, but we do anticipate that, well, maybe I shouldn't say even we anticipate. We're interested to see what happens when we perform that new training. We're hopeful that it will produce a better audio. Yeah. And is this, are you, this is a consumer electronic device. i mean that's your target market but it looks to me as though it would have traction in the music production uh industry is ever are you working on both sides or are you solely focused on um on consumer i think i think we we we want to make products that we can um use ourselves and really enjoy and and have you know no qualms with using them above any other competitor products in our potential price range or lives um products that we can recommend to our friends and family some of our friends and family just because of the nature of what we do you know they are music people they are music producers you know and but i think we're careful you know it can be very tempting to go too far down the road of building features that over complicate an experience if you're primarily thinking about producers.

47:19We really want the experience to be simple, elegant, and deep, but not so deep that we lose people along the way. We really want it to be like a six-year-old can pick this up and just just enjoy it right away and it becomes their go-to music device yeah uh the um i mean i'm just uh thinking you know i'm i'm getting old i'm hard of hearing uh it would be great to have a device that could could separate out one voice in in a party, for example. And, you know, that's beyond music. But on the business side, has the original STEM been profitable or is it still a sort of a development product and you're looking for to make profits on your next product?

48:22The original stem was actually more profitable than we expected after all the twists and turns with the partnership. So we were able to produce a net profit on those 96 ,000 units. And the gross profit was even better because the margins were quite good. I mean, without sharing too much, we were able to return enough to keep the business funded for many months and to work on new technologies, specifically in audio, but also in video. We applied some STEM separation techniques to video, and it got some really interesting results. And on our site, you can see the STEM projector product, which is like a handheld projector that will project on a wall or a ceiling, which also uses machine learning to augment video and transform it in a similar way that the STEM player does with music.

49:22So, you know, the business is in a position where I think we'll be going into the STEM to launch with a good amount of interest in the products. but yeah of course as a company that carries working capital and inventory it's never been the the kind of easiest business model to choose but that said is there any business model that's easy at the end of the day it's about you know you know making it something great producing a great product for customers communicating it well and it doesn't matter what your business model is you just gotta kind of pull your socks up and do it yeah that's fascinating uh okay well we're almost up to an hour uh is is uh what's the website that people can look for uh the website is stem.tech stem.tech yeah yeah i mean i've visited it but uh uh okay well this is really fascinating you're gonna stay focused on music uh or or as i said this technology it seems could go so many different places that's a great question um i think for us when i first um started the company i think one um goal that i had was to uh do as many different projects as possible almost to hedge my bets you know but right as as i've kind of become more um aware of uh just kind of the world and and how complicated projects come together especially ones involving many brilliant people i've become more appreciative of the fact that it's not all it's not always about like, you know, what, it's not even as much about what you're doing.