We're just days away from the Music Tectonics conference October 24th to 26th in Santa Monica and we have a very compelling three-part episode to get psyched, three topics critical to what's happening in music innovation right now.
We kick off things with a dive into streaming fraud, the billion-dollar problem laid out by Beatdapp's co-founder, Morgan Hayduk. Then we examine the seismic evolutions in music data and data transparency that's transforming how rights holders are getting paid with Serona Elton from the Mechanical Licensing Collective. Listen through to the end of this episode to hear master producer Daniel Rowland talk about his experiments with all the major generative AI music models released in recent days and discover what sounds good, what sounds bizarre, and where generative music is heading.
Looking for Rock Paper Scanner, the newsletter of music tech news curated by the Rock Paper Scissors PR team? Subscribe here to get it in your inbox every Friday!
Join the Music Tectonics team and top music innovators by the beach for the best music tech event of the year:
Listen to the full episode here on this page, or wherever you pod your favorite casts.
Listen wherever you pod your casts:
Episode Transcript
Machine transcribed
0:00:10 - Dmitri
Welcome back to Music Tectonics, where we go beneath the surface of music and tech. Dmitri here and we're just days away from the Music Tectonics conference October 24th to 26th in Santa Monica and we have a very compelling three-part episode to get your head ready for the most cutting-edge topics that will be on the minds and the agenda this year, three topics that are critical to what's happening in music innovation right now. That's the thing about this podcast and our conference we're agile enough to give you info that keeps you on your toes of changes coming through right now. Today, our three guests are going to hit on streaming fraud, the billion-dollar problem laid out by Beatdapp's co-founder, Morgan Hayduk. Then we have a segment on the seismic evolutions in music data and data transparency that's transforming how rights holders are getting paid more of what they have coming to them. That's with Serona Elton from the Mechanical Licensing Collective. Listen through to the end of this episode when we have Daniel Rowland talk about his experiments with all the major AI-generative music releases of recent days. A shortcut to make sure you understand the differences and nuances between them all.
That's a packed episode meant to give you FOMO. Sorry, not sorry. Let's go straight to my interview with Morgan of Beat Dapp about streaming fraud. At Music Tech Tonics, we focus on innovation of all stripes in the music industry, from creator tools to artificial intelligence, from the metaverse to music and gaming and fitness, but we're not a flavor of the month community. There's real and tough problems to be solved in the business of music. Innovative founders and execs that join us on the podcast and at the conference use innovation to help solve these problems. My next guest is a shining star in this realm. Morgan Hayduk is co-CEO and co-founder of.
Beat Dapp, a fraud detection company for the music industry. Morgan, welcome to the show. Thanks for having me. Dimitri, it's a pleasure to be here. I'm excited to see you in a week at Music Tech Tonics. You're speaking on a panel and we'll get into a little bit of that topic now, but let's just dive in. Why has streaming fraud become such a big topic lately?
0:02:20 - Morgan
It's a great question. It's one of my favorites because it's a tale that extends beyond the music industry. You can think about big pots of money, of revenue that are accessible on the internet as wonderful things, amazing innovation that facilitates e-commerce and trade and financial institutions to do all the things that they do, but it also becomes a vector for attack for motivated, often sophisticated and coordinated bad actors who are looking to reap some of that reward through oftentimes very sophisticated tactics. Music and the emergence of streaming since 2006, but really over the last 10 years, the way that the revenue pool has accelerated, the way that the consumption of streamed audio has just exploded, has made music, I think, a new attacking ground for fraudsters.
It's really, I think, a fairly simple reason to understand why they've targeted us. You have consumer-facing platforms that always have a trade-off between security and growth, with a direct revenue payment outcome tied to consuming audio time, to listening to music. When they look at users and platform dynamics themselves, and the ability to recoup their investment fairly quickly or clean money that they're looking to move from dirty money to clean, music offers an opportunity to do just that. It's not unique to us, I think, while we feel it specifically in the industry and are talking more about it today. This has been we're executing a playbook that is tried, tested and true across other verticals to increase security, increase the integrity of the product for everyone who makes money in the industry and ultimately, hopefully, push fraudsters somewhere else.
0:04:00 - Dmitri
Interesting. How big of a problem was this? Then maybe you could tell us a little bit about what does it actually look like? What does streaming fraud look like?
0:04:08 - Morgan
The problem. We estimate it between 5% and 10% of all streams. If you think of each individual percentage point of global market share and streaming is worth about $175 million it's not hard to get to very big numbers very quickly. The reality is the way we view it is at 1%, it's a problem. At 10%, it's stop everything. It's the biggest problem we have. We're confident that it lands somewhere in that range. We're not alone in this. Other DSPs have said publicly 5%, 7%, 10% themselves is what they see on their platforms. Obviously, if that's a reported number, it may actually be higher, because oftentimes that's what they're catching, not what they're entirely having happen on the platform. Industry analysts I think JPMorgan, Goldman Sachs have also put 5% to 10% in reports that they've written. We feel very confident, not just in our own data and what we've analyzed, but also what the other industry has triangulated other players in the industry have triangulated to as the range. It's a startling range.
0:05:10 - Dmitri
So what do we mean by? What does it look like? What does streaming fraud look like?
0:05:15 - Morgan
It looks like bots or stolen accounts working in coordination to drive, for the most part, revenue to non-music actors in the industry. So what I mean by that specifically is that what we see about 80%, maybe a little bit more, of the activity we see is not targeting what you and I think of as chart music, as popular music, as the stuff that everyone is listening to on Spotify or Apple or Amazon or any of the other platforms. It's noise, it's poor, low fidelity ripped audio from other places. It's the things that we really aren't finding in our day-to-day consumption but has been put onto these platforms specifically to target with streams to recoup that revenue. And so about 80% of what we see in that bucket and it's, I say, bots or stolen accounts, and oftentimes stolen accounts that are run as a bot farm. But there is a difference. A bot to us is an account that's been specifically set up only for the purpose of streaming music and targeting music for fraud. Stolen accounts are probably the more pernicious and honestly effective not to give fraudsters a roadmap tactic, because you're able to hide your intentions among this sort of natural, organic listening behavior of the owner of that account.
I usually pause here because I think stolen accounts is one of these things that, like, we all heard the stories. We all heard about data breaches, experience how to data breach Facebook's had a data breach. Everyone's had a data breach over the years. What I don't think the average consumer appreciates is that something like 90% of login attempts on the login wall of most consumer facing websites are what's called credential stuffing attacks, and that is where someone has obtained login credentials from another site's breach and is testing it on other platforms. And so every time there's a breach, it's not just the platform that was impacted, it's the ability to go and test those credentials everywhere else. So the ability to obtain stolen accounts is easier than ever. It's basically like my co-founder, andrew, often says it's like spinning up EC2 instances on Amazon. You just have to tell someone how much you need and for how long, and they can get you the volume.
0:07:15 - Dmitri
Wow, so you could be logging in to Spotify with Facebook, logging into something else with Facebook, and then some third party not even Spotify or Facebook could somehow have a breach and then all of a sudden, people get access to your Spotify account, for example. Exactly, right.
0:07:32 - Morgan
And look, I'm guilty of this. I probably shouldn't say it, but password hygiene is probably not on my list of accolades, and I think I probably speak for a lot of folks when we say that. And then, when you're thinking about the sites that you care most about, protecting those login credentials for, I would guess your streaming service password and username are not at the top of that list, and so, even if you're great about it for your bank, you're probably not resetting that password or using something truly unique and hard to crack for other platforms, and when data breaches happen, that's the hope of these fraudsters.
0:08:03 - Dmitri
So how does this impact artists and labels, this whole situation with streaming fraud? Yeah, I mean.
0:08:08 - Morgan
I think it's painful for everyone, but particularly for artists, because, one, they're probably the furthest removed from being able to have any impact on what's happening at a sort of platform security level. And then, second, the revenue flows through to them proportionally, and so if you're moving market share by any number of points at the aggregate, you're having that flow through directly and consequently and proportionally to all of the artists, big and small, who are making a living in the music business, and I think that's to me that's the sort of animating motivation for the company. It's like there are a lot of people who are not empowered to do anything about this but are on the receiving end of bad outcomes, and we want to make sure that if you're working incredibly hard and doing all the things you have to do to break through and to have a career in music, you aren't being punished by actions not of your own, but of just the susceptibility of the industry to being defrauded.
0:09:05 - Dmitri
Gotcha All right. So that's a good context of sort of how big it is. What is it, how it's impacting the industry, what exactly has beat that and how are you fighting streaming fraud?
0:09:15 - Morgan
We are a venture backed startup.
We identified this issue some number of years ago through conversations with platforms, with labels, with artists, with lawyers, anyone and everyone who would sort of speak about it and recognize that there was clearly a gap in the market for a third party who could come in and solve this problem, not just on a one by one basis, but to the benefit of everyone by working together.
And I don't mean that in some sort of fluffy, we can all work together, kumbaya kind of way.
I mean, literally, this is a big data problem, and the way that you get ahead of very sophisticated and clever fraudsters is by having the biggest and best data set to analyze across platforms and understand what tactics are being used and how to spot those patterns on a platform by platform basis, but informed by what's happening across all of them.
And I'll just give one sort of specific example here when you work with a smaller platform, it costs far less to analyze all of their data and look for really sort of unique edge cases, and so one of the advantages of being a company that looks across platforms of all shapes and sizes is we can test and develop models on small data sets, looking for edge cases that you probably would never look for in the massive sort of Herculean data sets of some of the bigger players, but then apply that learning to them anyway, and so there's a really natural lane in any industry impacted by fraud for trusted, neutral third parties to come in and look across a competitive data set like. DSPs are not going to share directly with one another their most granular platform data, so you need someone to sit in the middle, be a trusted intermediary and be able to provide that benefit back to everyone.
0:10:50 - Dmitri
So who are your clients?
0:10:52 - Morgan
That is a great question that we typically do not answer on podcasts. I'm sorry.
0:10:56 - Dmitri
Morgan, I don't mean who are your specific clients, I mean who are your types of clients.
0:11:00 - Morgan
Yeah, we work with sort of four client types. We work with DSPs directly, we work with music labels, we work with distributors and we work with collection societies.
0:11:09 - Dmitri
Gotcha. Okay, all right, that makes sense. So a couple of different approaches, probably with those as well. So have you heard of specific artists or labels that have been affected by streaming fraud?
0:11:19 - Morgan
I mean all of them have been affected by streaming fraud, I think publicly.
A number of them have started speaking out about this topic, whether in the context of new economic models or in the context of things that we need to get our arms around sort of better as an industry.
But yeah, I probably don't go maybe not a day, but certainly not a week without hearing an anecdote from an artist or from a label, and some of the worst stories that I hear, the ones that sort of make me the most sad, particularly on the label side, is when someone comes in and says you know, I worked with a third party marketing firm that I was casting about on Google, looking for someone to help me with playlist strategy or someone to help me with artist marketing, and they promised me the world in terms of being able to drive streaming volume around the first week of a release or some new single, and so we paid them.
You know, sometimes, I think recently, I heard an artist, a small label, say they paid $25,000 to a firm who delivered them artificial streams that were then flagged by the DSP as fraud and resulted in their artist being demonetized on the platform. That's a horrible outcome and it's all too common because at the beginning we talked about the sort of 80% being financial fraud. The more painful for the industry part that's directly painful for artists and labels is the 20% that is. I thought I was buying third party playlist help and I accidentally bought ClickFarm and the gray in between those two things you know is all too common because when you search for help me get more streams on DSPX, you will be inundated with companies that are offering you services that are definitely fronts for ClickFrods and farms.
0:12:55 - Dmitri
I mean, is there ever a legitimate third party that can get you more streams or, or follows or things like that, or is it always this sort of like bought farm stuff.
0:13:08 - Morgan
I mean, I would offer a couple of thoughts. The first is you know, if it sounds too good to be true, it probably is so. If you're promised explicitly X number of streams for X dollars, you are more than likely buying ClickFarms and stolen accounts. If you are talking to and there are so many great vendors and so many great technology companies, marketing firms, agencies in the music business and outside the music business who want to break in, who do incredible marketing strategy will help you, you know, try to achieve virality on UGC platforms that spin into streams. There's lots of things that you can do that are legitimate. I don't want to like foreclose on the entire vertical of marketing out of fear that you're accidentally buying, you know, fraudulent streams, but if the, if the, if the cost benefit is presented to you explicitly in terms of you can buy this many streams from us, you should not use that vendor.
0:13:54 - Dmitri
Right. I know there's a lot of companies out there that are kind of like what we used to do at Rock Paper Scissors. We did PR for artists and labels and tours. Now there's kind of the PR version of that as playlists like curators whether they're institutional curators who are hired by the DSPs or UGC curators out there that are just making popular lo-fi chill playlists or whatever else it is. So I'm sure that stuff you know that varies as well about how successful they are, but it would make sense to me that you could. You know there's other marketing strategies besides bought farms.
0:14:31 - Morgan
Yeah, like there are lots of good ones. I mean, I mean I would do all of the other ones. Sometimes people say like what should we do? And I said anything else.
0:14:36 - Dmitri
Yeah, yeah, I mean, do artists and labels have anything they can do to help stop streaming fraud?
0:14:42 - Morgan
I mean for us, you know there are certain yeah, there are certain things that they can do, certainly from an artist educational standpoint, because this isn't typically a decision that's made at the sort of parent, sort of macro level. It's made in within an artist's camp or through management, and so it's just about sort of educating them that these things are out there and they shouldn't step on the rakes that are laying around that might get them in trouble. And then I think also, it's it's worth always just pushing for more accurate detection across the platforms. I mean, if I could ask anything of the artist community and the label community generally, it would be to say, hey, that can up, is enough, there's a proven roadmap here. It's worked in e-commerce, it's worked in financial industries.
I mean we have a great Canadian company called Verifin that we often analogize to, because they built this exact business in the financial services industry. They started with credit unions and small banks in Canada and they worked their way up to using, you know, being a vendor to JP Morgan and Wells Fargo, where they're looking at their data for fraud in bank transaction data and for the exact same reason. You know, jp Morgan and Wells aren't sharing with one another, but they will work with the trusted third party. So, you know, advocating for, you know, better technology, the application of that technology, that, to me, is the is probably the biggest ask I would have of the artist and label community, because the benefits really flow through, you know, ultimately in the end to them.
0:16:01 - Dmitri
Because I've largest share of the money is going to real artists and labels rather than box fans and white noise.
0:16:07 - Morgan
I don't know Exactly, Exactly, Honestly, like whale sounds. You know you've had a good run, but let's we love whale sounds.
0:16:13 - Dmitri
We love whale sounds. I don't want to pick on whale sounds.
0:16:14 - Morgan
I should. I should actually say there there's lots of white noise and sleep aids and all of these other things that are legitimate and it's funny because the patterns are clear in the data. You can tell when something is asleep aid versus when something is being played nonstop 24, seven, 31 second tracks by a cohort of users that is, you know, small and very purposeful in what they listen to.
0:16:36 - Dmitri
Yeah, at the music tech topics pre conference, mark Mulligan from media research said you know everyone, every six to 12 months everyone's up in arms that you know Spotify is getting delivered 60,000 tracks a day. No, wait, that's 100, wait, that's 120,000. He said it's going to be a million very soon and it's interesting to think about, like what percentage of those are such lightly created tracks that maybe aren't going to get a lot of listens, or maybe I mean that that sort of points to why there needs to be a technical or a technological solution for sort of, you know, just vetting like what's legit, let's what's legitimate out there, and then in a way you can. It's hard in some ways it's probably harder to to editorialize what is a legitimate track as genres become more diverse and you know asmr tracks could be up there and that's, everyone's got a podcast.
You know there's so many different possibilities that could go through. You know the DSPs were thought of as retailers but they are UGC in another sense. We're really in this weird in between zone and now musical creators are more UGC than studio production and some, you know, in some categories. So it does make a lot of sense to have some technical solution, but it's hard to vet on the front end it's. It's probably easier to vet the way you guys are, in the sense of looking at weird streaming activity.
0:18:02 - Morgan
I think that's exactly right. The, the democratization of access for creators that streaming has created for them, provided them, is one of the biggest positives and, I think, one of the most positive developments of the music industry over the last decade. Plus right, the discovering, you know, sort of bedroom pop stars, the Billy Eilish's and others of the world is awesome, and I don't think as an industry we want to put gates up at the entrance out of fear, because ultimately what will come as a consequence of that is a loss of some of the most creative, most talented people that we could be exposed to, and that would be a shame. So I agree, technical barriers to entry, to access are really challenging and also subjective, and I don't think you know one thing technology doesn't do well is to adjudicate the subjective preferences of people.
So our approach, which is to look at the data after consumption fast enough that obviously illegitimate activity won't flow through to a sales report, is really the right way of looking at this, because once you have even a couple of days, a day, a week worth of data on a particular user, a track and artist, you can make a lot of highly, highly confident predictions about the legitimacy of the consumption around that track or the behavior of that user.
And so our view is that within a day, within a couple of days, we can start making high confidence determinations that will inform things for our clients, like should this be on an algorithmically distributed playlist or not, should this be eligible for charting or not? And then, by the time the month is done, with the highest possible confidence, we can say exclude this from sales reports, it's fraud, keep this in its legit, and that, I think, is the best outcome both for creators on the front end to make sure that they're not penalized for uploading something new that we've never heard before. That's going to be great but also protects the integrity of the industry on the back end, before the money starts flowing through the pipes.
0:19:54 - Dmitri
Right, yeah. Well, you've really hit upon a really kind of a win win solution for the music industry win for you guys, but also for so many other players in the mix, maybe not the fraudsters. So, Morgan, your co-founder and co-CEO, Andrew Beatty, there at BeatDap, is going to be on a panel at Music Tech Tonics called the X billion dollar problem emerging solutions to the music industry's streaming fraud conundrum. We've also just confirmed Michael Palschinski, who was VP of strategy at SoundCloud during the moment when they did all that interesting artist centric payout stuff. Our chief strategy officer at Rock Paper, scissors Tristra, Newyear Yeager, will be moderating the panel with you. So we're finishing out that panel, but I'm curious from your perspective, what are you hoping to get out of the Music Tech Tonics conference next week?
0:20:41 - Morgan
I mean, first of all, I live in Toronto, so any excuse to come back to LA, I will take a Music Tech Tonics is one of the best ones. You know, it's this right. We have a message to share with the industry. We have a lot of folks that we would love to work with, to be advocating for, you know, the solutions we're developing. You bring together, as you know, a stunning room of awesome people who are influential in the industry, and so the opportunity to be there, to talk to them, to hear what's concerning them and hopefully do some business and also just build some advocates, is a win. Win for us and, like I said, I wouldn't miss a reason to come to Santa Monica.
0:21:16 - Dmitri
Amazing, great Morgan Hayduk, co CEO and co founder of Beat Dap. Great having you on the podcast. What's the best way to find out more about Beat Dap and get in touch?
0:21:25 - Morgan
Morgan@beatdap.com, or you can come in off our website, but feel free to drop me a line directly. Always happy to hear from folks who are listening to the pod and hope to chat again soon.
0:21:35 - Dmitri
Awesome, we'll see you on the beach. Thanks so much, morgan. Thanks, dimitri. That was a masterclass in streaming fraud, but I'm looking forward to learning even more when Morgan's co-founder, Andrew Beatty, joins the music tectonics panel next week. Called the billion dollar problem, along with boomies Adam Rabinowitz, music tech strategist, Mike Palschinski formerly with the MLC Cloud, and our own Tristra. New Year, jager.
Okay, time to switch gears. We're going to take a quick break and when we come back, tristra has a very, very compelling conversation with the MLC's, Serona Elton, with an update on how the industry is leaning into, the importance of getting rights data right and the role of transparency and getting royalties to the right people will be right back. You can still get in on the conversations like these. At the music tectonics conference. We extended our regular price tickets until noon Pacific time on Friday, October 20. Right now, tickets are still $350 for three days of mind melding and mingling with music innovators, but after October 20, you'd have to pay the walk up rate $450. So get over to musictectonics.com and grab your ticket. That's also where you can find out all the who, what, where for the conference October 24 to 26 in Santa Monica, california. Don't miss out. Okay, we are back, and now Tristra has that interview with the MLC's brilliant head of educational partnerships, Serona Elton. I got unusually excited about data after listening to their conversation. Over to you, tristra.
0:23:10 - Tristra
I am talking right now with Serona Elton, who is head of educational partnerships at the MLC. Thank you so much, Serona, for making the time my pleasure. Thanks for having me. So you have a lot of expertise and a lot of aspects of the music business, but I really wanted to talk a little bit today about one area of your wide ranging expertise, which is data and how the role it plays in the music industry. So we're at a really interesting time when it comes to data. It feels like some things have gotten way better, or some things are remaining challenges or maybe have gotten even more challenging. What's your take on it? How would you sum up where we are right now?
0:23:50 - Serona
Well, data is more important than ever before because the volume of transactions that we are dealing with, really in society as a whole, but especially in the music industry, the volume of transactions has just exploded, and you just don't stand a chance of getting those things, those transactions, which, in this context, I'm talking about royalty transactions money flowing, it doesn't flow correctly without the correct data, and so the role of data is more important than ever, both in terms of making sure that the right people get paid and making sure that people can discover and find what they're searching for. You know, we don't think of it as data when we go on to our favorite digital music service and look for, you know, the latest track by our favorite artists. But as soon as you start typing in that name, you're actually using data and that's how you find them, instead of flipping through bins of records at stores. And so data is critical for discovery and data is critical to get people paid.
0:24:57 - Tristra
What has gotten better, to your mind? Where? What improvements have we made like as an industry? Where has, where has there been real progress when it comes to either generating better data, handling it and connecting all the loose threads? What do you think?
0:25:12 - Serona
Well, I think we have seen a couple of really important improvements. I think people's awareness of data and what we mean by that whether that's people working at music companies or whether it's creators themselves, songwriters, recording artists I think there's way more data awareness than I've ever seen before in the industry and, you know, I frequently come across people using terminology that really used to be only used by a small number of people that did nothing but data work. Now it's, you know, rolling off the tongue much more commonly, from all kinds of different people involved in different sectors and at different stages in the process in the music business. So I think we really have an amazing explosion of awareness right now, which is amazing. And I think our data practices have improved a lot in terms of changes in processes, where, when you can't like get to the next step, you know do not pass go unless you have provided certain critical information. We're starting to see that happen more and more throughout that process that goes from creation right through to distribution and then royalties flowing. We're seeing more checkpoints put in place where people must provide data in order to continue down the path, and that leads to better data practices and capturing right. It forces your hand, and so I'm seeing more of that than ever before as well.
I think where we've I don't want to say we've gone backwards, because I don't think we have I think we've made amazing progress, moving forward, but we just have more data than ever before.
The volumes are tremendous, and so trying to keep up with you know the, the ongoing educational effort I mean probably in the time you and I are talking here, there'll be, you know, several, more probably tens, if not hundreds, of people who will have just written their first song or made their first recording, and now they need educating.
And we're never done right. You know, even if we all of a sudden wave to wand and educated everybody who's doing this already, tomorrow there'll be a whole bunch of new people that need that education. So, you know, I think keeping up with the volume of music being created and the volume of new creators coming into the marketplace continues to be challenging. And I think, lastly, we've seen a lot of major moves in terms of data transparency, and that you know that that shows itself in a number of ways, whether that's organizations like the MLC making their database publicly available, whether it's more digital music services carrying credits, information and making that visible to users of the platform right fans of the music. We're seeing more data transparency than ever before as well, which is very encouraging.
0:28:13 - Tristra
And a lot of that has benefited songwriters, one could argue, or the rights holders that are related to compositions. I mean, if I'm just thinking of, I mean not to not to play favorites here, but Spotify is recent announcement of, I believe, that songwriters could either promote themselves or there were special playlists around songwriters, and I know Apple music just added a bunch of new credits and credits and new places, so it's kind of interesting and exciting. It's like something that I know many people in the industry were asking for and demanding for years and it's cool to see it all come together. It'll be interesting to see where it takes music fans.
0:28:48 - Serona
Absolutely. I mean, the kind of discovery that that sort of visibility enables is mind blowing. You know you could be looking at the credits and say, wow, this person you know wrote or co wrote this song. Let me see all the other songs you know, let me listen to recordings of songs that this person's also written and you might discover some amazing new music. You know you might say, wow, I've got a drummer on this track is just fantastic. What other tracks as this drummer you know perform drums on and then discover you know yet more music. So the discovery potential through the visibility of credits is really just so exciting and I feel like we're just at the early days of that.
0:29:29 - Tristra
So we talked a bit about some of these, the technology that might enable that, like, for instance, having credits and being able to click, click through and really find, or tap through and find, find your next favorite track based on I don't know the fact that Manu Kache played on it, or something like that. I was just thinking of my own example. Like you know, a lot of people my age I'm dating myself got really into jazz and certain international music thanks to Peter Gabriel and the fact that you could kind of you had to manually trace it, but you could find all the session musicians that he brought on that were just incredible performers from around the world, and I think it'll just really like these. These innovations you've just been mentioning are going to really superpower that kind of discovery, especially in, you know, in younger music fans who are hungry for new things or for finding, going down rabbit holes.
0:30:22 - Serona
Absolutely. You can just imagine the fun you can have, and it's really important that fans of music really demand that we get there. You know, because we're only in early days, it's all technically possible. It's all a question of capturing the data and having it be connected, and that will take an investment of time and energy and software, and for companies to spend that kind of money, put in that kind of investment to enable that, they need to know that the users of those platforms really want it, and so so I encourage anybody who is passionate about that you know, post about that. Call out you know Spotify, apple Music, whoever and say this is what we want. Make it happen.
0:31:08 - Tristra
I love it. I love that. Fans clamoring for more information about who in the world was played congas on that track so wonderful, that makes my heart go pitter-pat. I love it. Okay, so behind every data point there's a lot of technology, but there's also a whole group of humans right. Every data point is also a human moment and in some ways, some of our challenges in the music business remain deeply human problems, or I mean challenges, issues, questions, how you know, how are you seeing that evolve? Like you mentioned, a lot more data awareness on the part of artists and rights holders and even, you know, listeners who are starting to really explore, based on the data they know is available, what human problems remain.
0:31:56 - Serona
Yeah, so whenever you look at you know how something happens, right, how, how anything happens. There's generally three things to consider. There's people involved in taking action, there's processes and their systems right, and so the people. Part of this comes down to. You know who is the right person that actually knows the correct information. When do they know that and how can they capture that information and what's their incentive to do that right? And so you can. You know technology.
The data problems we're talking about the music industry do not exist because we just don't have the technology. You know, I mean, I've joked for a long time. You know, historically, if you wanted me to associate songs and sound recordings, you know, give me, give me an Excel file with two columns and unique identifiers for each, and I'll put them in Excel. Right, it's not a technology problem, it's a process and a people problem, and that has to do with you know what is an efficient way to capture that information from the right people at the right time. That's been, you know, something. We haven't solved particularly well, but we're moving much closer to it. One of the things that excites me is I'm seeing an evolution of a number of tools that are designed to be used in the recording studio. There's a couple of them out there, and they're designed to make it very easy for all of those creators who are in that recording studio and collaborating.
And, let's be honest, so much music today is actually written by the writers during the recording process. Right, it does happen sometimes that people might get together and have a writing session and then go in the studio later to record it, but a lot of recording is happening while the writing is being done right there in the studio, and so the ability to make it super easy, super efficient to capture in a system, while everybody is right there in the studio, who is doing what you know, who co-wrote this song, who is playing guitar, is really amazing. Having that go from happening in the real world, in real life, right IRL into a system and then have that data be transported from that studio to the music publisher, to the record label, to the distributor and ultimately to the digital music service and to all of the kinds of rights organizations that have to pay people, that is super exciting. And having that capturing of data integrated into the creative processes is where I think we're making major progress, and the key there is that the people in those studio sessions need to realize how important it is and not leave that session until that information is captured right. That's a key people change that can happen.
I think staff members at Record Labels and Music Publishers are really well aware of the importance of data, and you can't really not have a record company or a publisher today without a team of people dedicated just to setting up your data about your songs and about your recordings. But even then, you know, there's always a battle there has been as long as I've been in the business to get more funding for staff to do more data management right. So that's always a, you know, a battle to be raged inside of those companies. And digital services need to. You know are run by people. People make decisions right Decisions to add text, add credits to what you can see on a digital service. That's a decision made by somebody at a digital music service, and so we just have to reach the decision makers who get to decide what money is spent on and influence them about how important it is.
0:36:06 - Tristra
I want to switch gears just slightly, but this is still on the topic of humans attempting to wrangle data, and one thing I think is really exciting about what the MLC has done and you mentioned transparency earlier is to sort of create these tools, these portals, where any rights holder or any songwriter or, you know, independent publisher, anybody could go and look for works that had maybe not had fallen through the cracks, hadn't been fully claimed, maybe the data wasn't correct, maybe it was missing, but I really love that we've reached this. We've sort of crossed a rubicon of transparency, right.
It feels like even 10 years ago and you can speak to this better. You've forgotten more than I've known about this. I think you know we're looking back. Everyone had their database and that was their thing, and it was siloed or proprietary or only available you know, sort of in a commercial kind of setting, whereas what y'all have done is bring so many things you know and make them accessible to people, just so they can look and manage their own data in an environment that makes you know, makes human sense, right, instead of in some huge spreadsheet or some other format that's very difficult for most humans to read. So that's a really exciting decision that was made. How are you seeing that decision playing out? Any thoughts on that?
0:37:27 - Serona
Oh, we're super excited about it and I do feel, by the way, I should apologize for my scratchy voice today. Too many events going on. Too much talking about the MLC events.
0:37:37 - Tristra
It's hard to educate without talking. No, it's very hard.
0:37:41 - Serona
It's my favorite topic. No, we're super excited, and so you know. I will firstly give credit to the Music Modernization Act that changed the law that led to the MLC being created, because it mandated rightly so, and songwriters fought hard for this. It mandated that the MLC have a publicly available database of musical works ownership information, in other words, song ownership information. We're very proud to do that and that really, you know, we're not the first to ever have data available on our website, you know, as cap and BMI and other PROs have done that as well. But I think where we've really been able to take it to the next level is that our entire database is available for people to take a data feed. Now, that's not something that, you know, just your average person would want to do. It's a database, but what that means is that other companies you know that have the technology capabilities could take a feed of the entire data set right and compare that to their own data, do different things with it, you know, create different businesses that might provide value to, you know, everybody in the value chain. So that's exciting. But what we're extra, extra proud of are the tools that we built for our members to use, and one of the ones that you were just talking about had to do with making it visible to our members where we are still like, where we have royalties that we have not yet been able to pay out Because we have not been able to do something called matching.
So matching is when we are able to match a sound recording to the underlying song or musical work that it was based on, right, that connection, that link between the sound recording and that song. We call that matching. And when we have not been able to match, I mean we have computers doing it, you know machine learning algorithms, an army of human beings. But there's still some items that remain unmatched. And we are very excited to have launched this thing called our matching tool and what it does, and we really describe it as sort of illuminating the black box of royalties. Illuminating the black box because any one of our members can go into this matching tool, search it to find sound recordings of their songs and then recommend those matches to us.
Right, and it doesn't matter how much or little money we're talking about.
They could find something that literally had two streams, right, and has a fraction of a penny, but say, hey, mlc, please take the time match that sound recording to my musical work and pay me that money. It's my money and so having the ability for people to search this data set of unmatched recording usage and even in most cases they can click on a little play button and listen to that recording, because sometimes you know you have a common song title, it's going to be difficult to confirm if that's actually recording of that song, but this tool even enables them to listen to it. So you know it takes everybody playing their part to get everybody paid. And you know we think this tool, this matching tool, has so much value to the people, especially self-administered songwriters who are handling their own data and you know, again, it may not be a ginormous amount of streams, it may not be a ginormous amount of money, but it's their money and this tool lets them take proactive action to make sure that they get paid.
0:41:29 - Tristra
And it's cool that once they've claimed it and once the match has happened, it's there right, and so future royalties will flow to them, so that's really super exciting.
So about one last question that I've been pondering and you have a double role, or maybe even more than a double role, but in addition to your work with the MLC, you're also a professor and you're you know. This is such an interesting time for university education. There's a lot going on and students are facing a lot of challenges, but there's a lot of really neat, cool developments and all sorts of interesting things happening From your perspective. What does it look like right now? What is the next generation of music professionals interested in? What issues are you seeing coming up, or what exciting opportunities you know? What do you think about as a teacher when you go into the classroom and start working with these future music professionals?
0:42:25 - Serona
Yeah, it is a very exciting time. I think what we're seeing, certainly at the school where I teach at the University of Miami Frost School of Music, and probably across all of our peer institutions, is that students realize that it's going to take more than simply knowing their craft and really being amazing at their craft.
They also need to understand how to get a career launched and how to have a successful career Right. So you know, in days when you know you might show up and just learn how to play you know your harmonica. I always use that instrument because it doesn't offend anybody. I'm picking on somebody.
0:43:06 - Tristra
The anti-harmonica folks won't cancel you. They won't cancel me because you pretty much can't major in harmonica, really anyway.
0:43:12 - Serona
So you know a triangle or a kazoo or whatever like, pick your instrument. You can't just show up and learn how to be a virtuoso on that instrument. And then magically think a career is going to happen.
0:43:24 - Serona
You know there's so much more that you need to know, and so what you really see at music schools everywhere is a real acknowledgement of that, and you see courses in the curriculum that are focused on, you know, explaining the different parts of the music industry to the different students, depending on which direction they're heading. I mean, it's one big music industry, but there are some nuances about the way things work. If you're trying to, you know, make a living as a classical musician or a popular music musician or composer yeah exactly so.
So it will be, of course, specific to the particular discipline they're studying. But you see, curriculum having been adapted over you know the last probably 20 years to add in more courses that talk about career preparation, understanding the business aspects of your field, and a lot more practicum type experiences, a lot more hands on. You know internships or you know residencies, doing things where you're learning by doing a lot less sort of sitting and listening to somebody talk at you and a lot more doing and getting.
That you know invaluable feedback, and so I think you know it's very exciting in higher ed right now when you look across and you know some schools are doing it better than others and it does vary, but I think savvy students these days expect that, and when they come in and they don't just, you know, say wow, who's my teacher they say what does the curriculum look like? How are you going to be preparing me for my career? And that's a perfect question to be asking, and I think schools are really stepped up to be able to answer that in a variety of ways.
0:45:07 - Tristra
Yeah, I can say that I've witnessed something similar here at the Dickup School of Music at Indiana University. Even folks who are deep in the classical tradition are really trying to think outside the box and understand how they could play more and, you know, have a more fulfilling or well-rounded career through what used to be more nontraditional interaction with the music industry, and it's really neat to see there's so many interesting ideas and it's fun hanging around with young people. It inspires you and makes you. You know, involuntary optimism is the mood that I use. I like to, is how I like to sum it up, so it's great.
Well thank you so much, Serona, for making time and for sharing your amazing insights.
0:45:54 - Dmitri
See, I love how interesting Serona makes the importance of data for tracking songs and recordings. You can see why she's also a professor Nice one, tristra. Let's go to a quick update from Shaili about the conference, and then this segment you have to check out where Daniel Roland of Lander tells Tristra what he found out when he tested all the AI generative music systems that have been rolling out recently. Be right back.
0:46:18 - Shayli
Hey, music Tech Tonics besties Shaili here with some exciting programming announcements as we get ready to be back on the beach in Santa Monica for the fifth annual Music Tech Tonics conference. First, I'd like to highlight a new panel I've been working on called Transmedia Approaches to Music the intersection of Hollywood, gaming and music. The idea of this panel will be to dive into all the different opportunities for artists to monetize their music and build themselves as a brand across different medias. Sitting on this panel will be Brandon Bauman, global head of original content at Spotify, brooke Raskoff, global music marketing and streaming lead at Riot Games and an investor, to round out the perspective. Next, I'd like to spotlight Tatiana Cirassano's fireside chat, which will be rounding out the last day of the conference with high level insights that wrap together all the themes of Music Tech Tonics and projections for the future of music and tech drawn from midia research's deep data insights. Another panel I'd like to highlight, with some high level industry leaders, is the state of creativity in music and tech, with Matt Heninger of Moises, danny Dachacho of Splice, daniel Rowland of Lander, and this will be moderated by brilliant Danny Deal at BandLab.
Now, before we get back to the episode, I wanted to be sure to let you know about this cool new opportunity that my team just launched, taking place on the second day of Music Tech Tonics. We will have an AI Innovation House with a chance for AI companies to get demo spots and branding at the conference. Music Tech Tonics is always keeping on the pulse of trending tech. If you want to be a sponsor of the AI Innovation House, feel free to reach out to me at shaley at rockpapersizzards.biz, and we can schedule some time to chat. If you haven't secured your spot to the conference yet, go to musictechtonicscom and buy your badge now. Alright, back to the episode.
0:48:20 - Dmitri
Okay, if you come to Music Tech Tonics, you know Daniel Rowland at Lander. In fact, he's speaking about the state of music creativity next week with Matt Henninger of Moises AI, danny Dachacho of Splice and Danny Deal of BandLab. What a lineup Back to Tristra for this segment on generative AI music tools that have come out recently and experiments with them by Lander's Daniel Rowland.
0:48:43 - Tristra
Thank you so much, Daniel, for volunteering to join me. What I really would love to talk to you about today is some of the explorations you've been doing and sharing on everybody's favorite social media platform, LinkedIn. You kind of road tested a bunch of the larger models that are generating music, so from places like Meta and Google and a couple of other big tech places. I was just curious how those experiments went. What were your impressions as someone who's worked with audio and been dealing with and loving music for a long time? Let's start by letting folks know what are some of the bigger models that are out there that the public can play around with.
0:49:33 - Daniel
Yeah, so the ones that are readily accessible, a lot of people have probably heard and you mentioned a few of them already the Meta and Google offerings. Musiclm and MusicGen are two, musicgen being part of Meta's audio craft offering, which is a couple of different things. There's generative sound effects, so they have a model trained on sound effects libraries and also when trained on music. Those are two that I had done a fair bit of testing with. Since those were released, there's been other ones that have come out as well. Of course. There's a number of them to varying degrees of quality. The most recent one is probably from Stability AI, the Stable Diffusion, one of the image generators, and that's called Stable Audio. That's come out recently. They have a team internally, harman and I. That's a music division. They've had an open source thing and now they went out and got a data set from this company, audiosparks, which is a production music library. They trained on 800,000 tracks of theirs and they released an actual product that you can pay for called Stable Audio, which is a competitor to these other things we've been chatting about. Dash has one, too, and there's a number of them. I've tested them and the result is similar amongst all of them with some subtleties.
When I first tested the Google and Meta options, meta to me was superior, not just because of the output that I was getting, but because the input options with Meta and we've talked about this previously instead of just having to type in what I want, which is a very awkward way for people to engage with music for a number of reasons I'm sure we'll talk about you can actually provide an input file as an inspiration to direct where it's going, which is a little something people I think it's a little more digestible, feels a little bit more artistic to go about it that way, whereas with Google and a number of the other options, it's up to you to enunciate what you want to speak, what you want into existence.
That's why I really liked the Meta one the new and not to be too long-winded. But the new Stable Audio one from Stability is actually very good. I mean, the audio quality is superior to some of the other ones that are out there sonically, but it's still a bit of a crapshoot on what you're going to get, and a lot of it, like chat, gbt and other things like mid-journey and Dolly, it's reliant upon your ability to say what you want. We're still the music space so that everybody knows is kind of where the image generation space was three years ago, right?
Where people are saying it's not very good and it's never going to sound good or it's never going to look good, and we know that's not the case. Right, this tech improves. Look at the image generation space. That's still a little bit different than the music space. But don't expect to get something that's just perfect track back, because you're not going to get that. But what you do get is some stuff that can be very inspiring, whether it's you're asking for a fully composed track to sample or you're asking for samples and loops back. I want a jazz piano loop in the style of Nina Simone or whatever, and you're not going to get that back. But you're going to get something that could be useful if you have the abilities to take it to the next place and use it in the context of something you're already creating. And that's kind of where we're at with the technology.
0:52:47 - Tristra
So I would love to take a second here to unpack a little bit what we talk about when we talk about quality with generative AI results. So are we talking about? Is it true to the prompter, intention of the prompter, which I guess is hard to judge? It sounds like something we need to bring in a resident philosopher or something to get that in there. Music always, exactly, is it the audio quality itself? So how well it sounds, the resolution, and maybe you can give some better terms to that.
And then there's a problem and this is something I heard with some of the tracks you generated, which sounded kind of cool, but they were extremely incoherent. If we think about music as a language, right, it's like, as you said, it reminds me of some of the GPT models, like two or three. You could get back some really weird ass pros. That was quite inspiring. You're like, wow, this is really really weird. I never would have put those concepts or words together. So I'm wondering if you could speak for a minute about the different parameters by which you think we should be potentially judging these models and their outputs.
0:53:53 - Daniel
Well, I think you just said it very well which is one of which is fidelity, right? So the actual audio quality of it. Does it sound like a bad MP3? Does it sound like, you know, to give an analogy to kind of how we think about audio as engineers Is it a very lossy kind of compressed sound data-wise, or is it something that's for very high fidelity and right now, a lot of what you get back is has this very lo-fi sort of aesthetic to it, right, which can be a vibe, right? You could say that that's useful for things, I mean as engineers and producers.
A lot of times we go to great lengths to make things sound kind of grungy, but it's kind of you don't really have a choice right. It's going to kind of have that sort of a quality to it. And then, of course, the other side is, as you said, like is it accurate to what you were hoping to get? And quite often I mean it's going to be. You might have a component of what you are suggesting, right, like you want a synthwave track that kind of sounds like a weekend that has, you know, drums similar to whatever. You're going to get something like that back and as you continue to hit generate over and over again, you might get 10 things and three of those things might kind of be on point. That's kind of where we're at. So it's the. You know. Creatively, are you getting something back that fulfills what you wanted? And then, sonically, are you getting something back that suits the needs that you have? As an example, if I wanted to generate a track that I could take as it was and use in an advertisement, for example, when we're talking about generative audio specifically, not MIDI generation where you're then choosing samples and making it sound lush not so great for that at the moment, and it's similarly, are you getting back a track that sounds like it was composed by a human, if that you know, if that's the goal, you know it's hit or miss, but you can get some good stuff.
I mean, there's companies. I've certainly in my experience and from the experience, the experiments, excuse me, of other people heard things that are like okay, that's a track, now it might be a 20 second clip or a 30 second clip, maybe that works, and then it falls apart. Right, so you're still having to go and curate and curate to get that. But that's the type of things. These are the type of limitations just like we've seen in other areas that will fall away in the coming years.
And it's about it's not just about the improvement of the models. It's about the access of two of the companies building these models to the correct data sets to train these engines. That's one reason that I could be totally off base about this. But when you try to generate loops, individual instruments so I don't want a full production, I want a piano loop, a drum loop or, you know, vocal loop or whatever Very rarely do you get back something that's that's worth a crap, because I'm not so sure that the data sets that have been used to train these engines were loops and samples.
I think they were probably full tracks. Right, it's, that's you know, like.
Take the stability AI example. They trained on a production music library of completely finished, you know compositions and it's better at hitting that target to me than it is creating loops. So, again, we're starting to see deals come in place for companies. You know that are the content owners, are starting to make deals with the companies that are the creators of the technology, because you know they the content owners want to monetize what they have in various ways. They see the threat of generative AI and they know they need to kind of get involved with it or be opposed to it, and that's the tech side of things. They need that data to improve these engines. So it's fascinating to see kind of the back and forth and the deals are being done behind the scenes right now.
0:57:14 - Tristra
I mean, I could almost imagine interlocking layers of AI, so you'd have the generative layer that had been trained really well on a bunch of stems and Individual instruments or vocal lines, and then you'd have another AI model which would assemble the tracks in ways that were most likely to form a coherent, you know, composition or song or Soundscape. I mean, it could be really interesting, as I mean and maybe that's exactly what's gonna happen I think that has something similar happened with images. Yeah, it's it. That's a really, really interesting way to think about how we could build better AI.
0:57:50 - Daniel
Well, what you just said it would describe as the holy grail of all of this, right? Because, ultimately, what do I really want to get back? Right, do I want to get back a stereo mix that the AI has come up with, or do I want the AI to generate the vocal and the bass and the drums and all of these things intelligently relative to one another into a fully arranged composition and Provide me not just the mix but those stems back that I can then iterate upon and repurpose and what have you like? That is what, ultimately, where all this is going and what will be, you know, yeah kind of the ultimate version of this, at least in the foreseeable future.
0:58:25 - Tristra
Yeah, so it's. It's really interesting to think, too, about crafts, people such as yourself or artists such as yourself who can take, who can make more of these little bits and pieces that could potentially be generated AI. So one of the more interesting directions I've seen with some generated generative AI startups has been startups that really specialize in one or two types of sounds, so like drum hits or really specific kinds of synth patch Generations, like what what you know people used to pay you know folks for, back in like the 90s right, when you'd like right away to get like a patch from some custom patch designer To put in your in your synthesizer. Well, I think that there's some rule for some AI startups in that realm and I'm just wondering what you're seeing about some of these bits and pieces and other ways besides just creating a full-blown track that you're seeing generative AI poke its head up here and there.
0:59:23 - Daniel
Well, yeah, let's just kind of pull on the thread of what you, what you just said, which again, is pretty accurate. This is why Tristra's awesome.
0:59:29 - Tristra
Oh, it's all by accident. It's a problem. I'm a probability machine, just like AI.
0:59:37 - Daniel
So if we're trying to look at where AI is hitting the target the closest right, you kind of want to look at the most Simplistic use of it, and this is what you just said a minute ago, which is it's okay. So it's. It's obviously fairly difficult for AI to compose a three-minute song with a clear beginning, middle and ending. Okay, it's okay. Ai is not really great right now, generating Consistently. By the way, when I say it's not great, I don't mean that it can't do it, I mean it can. It can Repeatedly, consistently hit the target without you having to iterate through a million different things. Right, and that's where we're at. It can't do that, really.
So with samples and loops, well, loop, specifically, it's not there. But where people are actually doing quite well is with the generation of individual samples, or what we would call one shots. Right, and specifically for Non-pitched instruments like drums, right. So if you look at some of the stuff out there that's been productized In plugins right, so things that we would use inside of a DAW to create music with there's some quite good stuff. That's. It's advanced in that just in the past six months to a year pretty far. Companies like Audio lab, who make this thing called emergent drums right.
Which generates effectively a drum kit for you and you can choose to lock certain pieces of the kit. Maybe you like the kick drum but you don't like the snare drum, and then you can tell the plug-in, tell the model, do I, do I want to have a completely different snare drum? How much variance do I want from that existing drum to the next versions? And it will continue to iterate, iterate, iterate into until you like it. I mean, sony computer science labs has a product called drum net that does that as well. There's a few of those floating around, and I found that those have become quite good, right, they're actually getting to the point where they are usable in these kind of bite-sized generative. I use cases. So I think we're gonna start to see that, and that's been Again.
There's a number of products there I didn't even mention, but they're all effectively doing the same thing. And you know, you combine that with generative MIDI sequences, which we can do fairly well, right, because that's just about the positioning of notes. It's kind of the, you know, the sheet music that you use to play the sounds. That actually works fairly well. Now, can we then evolve that to loops and samples and then to the fully full tracks and have it be consistently artistically Fulfilling output. Yes, but that's kind of where things are headed interesting.
1:01:50 - Tristra
So one thing that we were talking about before we hit record here was if you think in music, if you're an audio thinker, it can be difficult to figure out how to translate all of those thoughts and feelings into text prompts.
In some ways, I think our brains are well, they're not wired, but our brains are more inclined to attach words, to visual elements. Right, either we have more practice telling each other about visual stuff rather than audio stuff, or I don't know, maybe there's something inherent in the way our brains are architected. But though I was just curious, if you've talked to other producers or other people experimenting with gender generative AI In music, you know how, how does it feel to try to translate the world you hear in your head into a few text prompts that you're hoping the machine will read properly?
1:02:42 - Daniel
Yeah, it's not exactly the most intuitive workflow for people who are A, for people who are used to composing and writing in their own music and B, I think, for people who aren't right. I don't think it really works for For anyone is the best way, the best interface for this sort of thing because, if you think about it, the realm of Really clearly being able to enunciate what you want from music is the realm of a pretty highly skilled producer, right Somebody who can?
pick the pieces they want and reference the things that they want and Really speak that into existence. That's not something the everyday person is walking around. I mean the way that Most people relate to music. It's either you know, okay, maybe you can't play an instrument, but you can. You can hum and go, do, do, do, do, do, do, do. I want something like that. Or you reference an existing artist or an existing song or something with. You know some content you've engaged with before. Those are, you know, even for songwriters are people sitting around jamming, right, that's a? Hey, give me a beat like a James Brown or give me a you know a baseline like so, and so that's how we typically speak about this type of stuff, if we're speaking at all about it, and it's all. Oftentimes people are just creating because they're physically engaged with an instrument. So it's a completely different Way of going about creation.
That text prompts doesn't really address so and I don't think it ever really will.
And the problem, of course, is With the legalities understandably involved in this. You can't just really say I want it to sound like Prince or something like that, because the data set Likely, hopefully, was not trained on Prince, because Prince is not authorized as catalog because the state hasn't to be trained, and those are the type of things that Some of. For this to really gain scale, some of those barriers will have to come down, and there's, of course, labels interested in this, and we see, you know, movement in that direction. But, yeah, so I don't have a solution, and a lot I've heard a lot of other people talk about this too that don't really have a solution about where this is going to go and what the interface Is it's going to look like. But we have to bring it closer to how people are used to engaging with and speaking about music for it to really, you know, become this ubiquitous technology that is integrated in all the products that we currently use and ones that you know we can't even proceed.
1:04:49 - Tristra
Yeah, there's something to be said for looking at how humanity has reacted to Audio or music for a long time. Right, like movement, dance, gesture and all of those kinds of controls and interfaces are still very much bleeding edge, right, and it's fun to think of the day where you could like, dance your song into being and it sounds very woo and I'm gonna get my patchouli out here and um, like you know, my chest and how would that become a drum loop instantaneously?
1:05:18 - Daniel
Yeah, we see this right, like I mean, even the new. This technology's been around for a while, but if you look at the new tiktok ripple thing right, which is where I can you know, like I did earlier, I can hum something and the AI will analyze that and attempt to build an arrangement based upon Did that input right? So it's not about me having to describe something. I can just you know, I can enunciate it or I can tap it on a table or I can, you know, maybe play it on an instrument and I'll get be able to iterate through and then type in oh I like what you did, but I wish this was more classical or it sounded more like Drake or that. Then you can start to iterate upon these Kind of more musical inputs. I think that's kind of a hybrid approach that it will start to see make its way into more professional tools, but right now it's not. It's in a very much kind of social media.
1:06:01 - Tristra
Yeah, I love that. I would love, I love the, the, the way you're talking about again these different layers that an AI could be baked into a bunch of them in different ways, but they would all sort of contribute to a very, very different experience of creating music, right, that would sort of take. Take us away from the you know what, the console that's been translated into a flat screen and Suddenly, you know a lot of more people can find a way to interact with this technology that couldn't do it before. Are there? This has like been so fascinating.
I love how you're talking about how this is getting incorporated into more professional products From social media. That's like so interesting, right, because a lot of what we've seen just With digital music and that it you know recently has been translating like the, the fancy studio console, into you know something like a very sophisticated DAW and then sort of simplifying that so people can use it and manipulate it who may not be experts. I'm wondering if you're seeing anything else out there, or or the just the first inklings of the way AI might help more people make music who may never have thought of playing around with audio before.
1:07:14 - Daniel
Yeah, sure, and I mean similar to what we've spoken about before. You kind of have to meet people where they are right. So, yes, I personally, as a professional, love using a DAW and there's even, you know, garage band and these other things that are still basically professional level tools. But you have to seek those things out, right, you have to say I want to make music, I want to download the software, I want to interface with it and there's, you know, some AI involved in there to kind of help me along my way and shepherd me to getting something that makes me inspired, it makes me want to come back and do that process all over again. But that is niche relative to to having tools like the TikTok tool I just mentioned. That meets people who aren't even thinking, maybe, about making music, but it kind of either prompted to or they see other people doing it and they can immediately engage with that. And that's why, you know, the social media component of music production and music making and collaboration, especially collaboration, I think is super key and we've seen that over the years with singing, right, when you know we want to do duets on Insta or TikTok, whatever right, snap, you see all that kind of stuff, but now having actual music production be involved in that is important, and there's been attempts at this over the years that haven't really caught on, so I think, but but they've always been ones that were AI wasn't really involved, right? So, okay, we give somebody a sequencer inside of their social media platform, but you still have to understand what, how that works, right, how to sequence a drum beat, how to? So just providing existing tools to a new market, I don't think works. I think you have to. The intelligence factor there is super important. So, yeah, I mean that's.
Social media is where we're going to see a lot of this, and also, to some degree, in the gaming community.
I think there's some really interesting things being done, whether it's in VR or you know things that are being released via, via, you know a gamified music experiences that kind of like. You know, rhythm games, of course, have been popular for a while, whether you look at Beat Saber or any of the other million of them that are out there at this point on mobile, desktop, console and VR. Like having those become places where you're not just listening to a track that you, you know, or maybe when you don't know, and hitting, you know triggers at various points, but you're actually creating as part of that with other people, super fascinating and where a lot of this is heading, where we're really trying to, like you know, allow people not just to participate but have this kind of bi-directional conversation where they're part of the creative process in environments that they're already used to, and that is that kind of malleable content, is where a lot of, I think, things are heading across. You know film, you know music, et cetera.
1:09:41 - Tristra
Yeah, it's going to be very interesting to tell my grandchildren about. You know, back in my day.
1:09:47 - Daniel
A song was always the same.
1:09:50 - Tristra
Well, I love it, though, because they will never have to suffer from their favorite song getting so ingrained in their brain that they can't hear it anymore, which is always a tragic thing for me, because you know. You know, when you play a song like so many times in a row because you love it and you want that same hit, that same high, and then, like the 10th time, you're like it's just not the same. Anyway, here's to, here's to, here's to um to uh, never having to lose that earworm feeling, um, for our future generations. Well, thank you so much, daniel. This has been a really fun whirlwind tour of your AI experiments and AI in general and how it's affecting music creation, and I can't wait to talk to you more at the Music Tech Tonics conference. Before you go, I would love to hear what you're looking forward to in the next coming weeks when we all meet up in Santa Monica.
1:10:46 - Tristra
What are you most excited about?
1:10:47 - Daniel
Oh, I mean, first and foremost, just getting to see everybody and hang out and learn right, I mean the tech tonics for me is like I feel like I'm I'm pretty or very well connected with kind of the future of the industry. But every time I go I learn about multiple things that I had no idea about and I hear you know, meet people that you know. I end up having relationships that take me in multiple directions, including on some of the stuff that I'm working on right now, that I wouldn't have had it. I not attended that. I don't want to be a pitch person for tech tonics, but it's such a well on conference.
It's ridiculous. It's in Santa Monica. What else could you want? No, I'm super excited I get to hang with you.
1:11:25 - Tristra
Come on, it's going to be fun. I can't wait. Well, thanks so much, daniel, for your time and your amazing thoughts, and I'll see you soon in Santa Monica.
1:11:34 - Daniel
Most definitely Thanks.
1:11:37 - Dmitri
Thanks for listening to Music Tech Tonics. If you like what you hear, please subscribe on your favorite podcast app. We have new episodes for you every week. Did you know we do free monthly online events that you are lovely podcast listeners can join? Find out more at MusicTechTonics.com and, while you're there, look for the latest about our annual conference and sign up for our newsletter to get updates. Everything we do explores the seismic shifts that shake up music and technology, the way the earth's tectonic plates cause quakes and make mountains. Connect with Music Tech Tonics on Twitter, instagram and LinkedIn. That's my favorite platform. Connect with me. Demetri Vica, if you can spell it, we'll be back again next week, if not sooner.
Let us know what you think! Tweet @MusicTectonics, find us on LinkedIn, Facebook and Instagram, or connect with podcast host Dmitri Vietze on LinkedIn, Twitter, and Facebook.
The Music Tectonics podcast goes beneath the surface of the music industry to explore how technology is changing the way business gets done. Weekly episodes include interviews with music tech movers & shakers, deep dives into seismic shifts, and more.
Opmerkingen