Have you ever wondered if AI can master my music? Maybe you're seeking answers to topics: the best AI mastering, is there an AI that can mix songs, is there AI for music production, can I AI make music, can AI make my song sound better, or maybe even is AI-generated music legal? Then check out EP 134 of the Inside The Mix podcast.
Join me for a riveting discussion with Bobby Owsinski, music production guru and educator, as we dissect the role of AI in today's music scene. Bobby demystifies the jargon and sheds light on how AI is shaping the way we create, refine, and promote music. From the potential of AI to serve as a digital muse to the transformative powers of automated audio mastering, we cover it all. His insights navigate us through the complexities of new-age music production, where AI meets creativity—providing foundational support while leaving room for the human touch to add depth and nuance.
Venturing further into the AI realm, Bobby and I tackle the notion of AI-generated music, examining its current limitations and possibilities. Can AI-composed melodies stand toe-to-toe with the intricate works of human artists? We grapple with issues from generic patterns to the emulation of complex genres and vocal licensing. Plus, we don't shy away from addressing the elephant in the room—copyright dilemmas in the AI era. For anyone curious about the future of AI in music and the legal labyrinth that it implies, this conversation offers an enlightening perspective.
As the episode comes to a crescendo, we explore AI's capacity to balance and refine tonal quality with tools like Gullfoss and Hornet's Sleek. While examining their potential to change the standard workflow in music production, we remain grounded, recognising the irreplaceable value of human expertise. As I say goodbye to Bobby, I encourage you to discover the breadth of his knowledge on his website, which is a goldmine for anyone looking to keep a finger on the pulse of music production's digital transformation. Tune in for a symphony of insights where technology and artistry harmonize in the evolving world of music.
Grab your FREETest Master at Synth Music Mastering TODAY! ✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸ Are you READY to enhance your music with my steadfast dedication to quality and personal touch? Bag your FREETest Master at Synth Music Mastering: https://www.synthmusicmastering.com/mastering
Buy me a COFFEE ✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸✸ If you like what I do, buy me a coffee so I can create more amazing content for you: https://www.buymeacoffee.com/marcjmatthews
Send a DM through IG @insidethemicpodcast Email me at marc@synthmusicmastering.com
Thanks for listening & happy producing!
Marc Matthews:
This is Steven Zero, and my favorite episode of In the Mix podcast is episode number 30, side chaining. Side chaining as a beginner is very difficult, but Mark makes some good points and tips to make it easy.
Bobby Owsinski:
You're listening to the Inside the Mix podcast with your host, mark Matthews.
Marc Matthews:
Hello and welcome to the Inside the Mix podcast. I'm Mark Matthews, your host, musician, producer and mix and mastering engineer. You've come to the right place if you want to know more about your favorite synth music artists, music engineering and production, song writing and the music industry. I've been writing, producing, mixing and mastering music for over 15 years and I want to share what I've learned with you. Hello, folks, and welcome to the Inside the Mix podcast. If you are a new listener, make sure you hit that follow button on your podcast Player of Choice. And to my returning listeners, a big welcome back.
Marc Matthews:
So at the point of recording this intro, I am just coming out the other side of the dreaded flu and sore throat, but I thought I'd cheer myself up and I went and bought myself a replacement new sub for the studio, so I'm excited to get that installed. I want to give a big shout out to Life Guitars in Exeter at the same time for it sorting me out last minute with some XLRs, as I needed some replacements for the studio on a Saturday and they were the only place in Exeter that I could find XLRs last minute and they did me a fantastic deal. So if you're ever in Exeter or Devon Southwest, go and check out Life Guitars. So, moving on to this episode Now, you'll find that when I start the episode, I am very excited and I still am listening back to this as well. When I've edited the episode, I had Bobby Ozynski on the show, so renowned music producer, author and educator, known for his expertise in music production, audio engineering and the music business. And you'll hear me say on the episode he hosts the Bobby Ozynski Xenocercle podcast and he's got a blog and does everything else in between. And I was so excited because, as I mentioned in the interview, he's probably my most referenced author when I was doing my masters and my undergrad degree with the Mix Engineers Handbook and I was so excited to have him on the show.
Marc Matthews:
So in this episode I took the opportunity to dive into the Musicians AI Handbook, which was released by Bobby, and it equips musicians and producers with the skills needed for success in the music industry and using AI. So Bobby talks about the rationale behind organizing the book in the particular way it's set out and how each section contributes to helping artists and producers navigate the complexities of AI technology in music production and also promotion. So we talk about opportunities AI presents for enhancing music production and how artists can leverage AI tools to elevate their compositions and arrangements. We talk about some common restrictions artists should be aware of and how they can navigate these limitations to ensure a seamless integration of AI in their creative process. We talk about myths and misconceptions surrounding AI's role in music production, and a little about marketing as well, and talk about some of the prevalent misconceptions that Bobby's encountered along the way, and we talk about a lot more in between as well with regards to AI in music production. So that's enough from me.
Marc Matthews:
Let's dive into my chat with Bobby Ozynski. Hey folks, and in this episode I am very excited to welcome my guest today. A huge influence on me as a mixing engineer, just in music in general, and I was saying off air, probably my most referenced author when I was doing my master's degree, bobby Ozynski. Hi, bobby, thank you for joining me today. I'm extremely excited about this. How are you today? Thank you for having me, mark.
Bobby Owsinski:
I'm pleased to be here.
Marc Matthews:
Fantastic. So I've got to do a shout out to Jay Gilbert of the your Morning Coffee podcast, who put me in touch with Bobby, and also go check that episode out as well. Episode 130. So, for those of you who aren't familiar, I've just got a brief bio here for Bobby. So he's a renowned music producer, author and educator, known for his expertise in music production, audio engineering and the music business, and he hosts the Bobby Ozynski's Inner Circle podcast. Do go check that out there will be a link in the episode description offering invaluable insights and interviews with industry experts, and as a creator of online courses and bestselling books like the Mixing Engineers Handbook. So I've got a copy of it right here.
Marc Matthews:
This is the third edition Fantastic book the Recording Engineers Handbook and, most recently, the Musicians AI Handbook. Bobby equips musicians and producers with the skills needed for success in the music industry, and today's chat is going to be about AI. So it leans into the recent book, the Musicians AI Handbook, and how artists can use it to enhance and promote their music using AI, as described by Bobby himself. So I thought it'd be a good place to start. Bobby would be with the book structure itself, so it's structured in three distinct sections. Can you talk a bit about those sections and explain sections rather, and explain the rationale behind organizing the book in that way?
Bobby Owsinski:
Well, the first section is primarily about the basics of AI, explaining all the buzzwords that we know, that we hear and sometimes confuse.
Bobby Owsinski:
Sometimes we mean one thing and it's actually another, and just in terms of what the you know, the normal terms that we hear.
Bobby Owsinski:
The second part of that would be AI copyright, which is very blurry and nebulous right now, and it's clearing up a little bit, but in fact, there's a lot of things that you have to watch out for when you are using an AI platform of any type.
Bobby Owsinski:
Second section is all about using AI and audio, and that includes generating AI music, using it music production, using it just for audio tasks like EQ, limiting things like that, compression, source separation, which it's very good at, noise reduction it's extremely good at that, getting better all the time. And then the third section is all about using it for promotion, because certainly it's very good at generating music videos. It's very good at generating copy for, for instance, a bio for you or maybe a release strategy or anything like that, and that graphics is, you know, exceptionally good at that and getting better all the time. So that's the three different sections, and again, the reason why is I didn't want one section to blur into another one. So I kind of defined the intersection between them a little bit, just so it's easier to get through the book.
Marc Matthews:
And I think that's a really good way to separate it. And particularly what I really like about the book is going through it. It's not you're all. You're describing and educating the reader on these different facets of AI, but you're also giving sort of steps, for example, with the marketing side of things. You're saying, okay, well, this, these are the chatbots that you can use chat, gbt, bard, which is now Gemini, I believe but you're also giving okay, so this is a breakdown of the prompts that you can use when you're using this particular for platform, and also with the, with the VSD plugins and those bits and pieces as well. So I think that's really really useful. Actually, okay, so this is what you can use, but here's how you can actually put it into action.
Marc Matthews:
Going back to the first chapter, and because quite the big advocate of AI in itself, are there any? Is there like a particular part of AI that if you were new to it, you should get started with? Because there's things like neural networks, there's large language models and deep learning. Is there like if you really wanted to understand how it's all put together? Is there a particular part to start with, do you think?
Bobby Owsinski:
Honestly, the whole idea behind that chapter was to bust some myths. Myths, primarily, and it's mostly like okay, there's a difference between a large language model and a neural network, for instance, and there are many times when a lot of these terms are used interchangeably and they shouldn't be so. It's mostly like let me put this into layman's terms so you don't have to be a computer engineer, computer science major to get through it.
Bobby Owsinski:
And just understand that there are differences and there's a history here. This isn't something that just started last year. In fact, it's been going since the 1950s the theory behind artificial intelligence. So, just to kind of break it down, it's like, okay, don't be overwhelmed by this, because it's not as overwarning, it's not as daunting as you may believe and again, I was trying to put it in terms where it doesn't feel like it's daunting or it's like, oh yeah, I can understand it, oh, I get that. So that was the whole purpose behind that. Ultimately, what's going to happen is this will all somewhat blend into the background, in that it's beginning to happen already, where we're seeing that things are happening so automatically for us when it comes to using any kind of AI platform that we're not even thinking about those buzzwords so much as we were in the beginning.
Marc Matthews:
Yeah, and I don't know the audience listening. I don't know about yourself, but now whenever I look for new platforms, I think one of the first things I generally look for now is how they incorporate AI. Maybe it's because I just love AI and artificial intelligence and everything that comes with it, or there are caveats involved with that with regards to copyright and everything else. And, as you say, we're sort of reading the book as well. You say that different parts of the world in terms of the copyright law is it does fluctuate, but I'll let the readers go into that when they read the book.
Marc Matthews:
But I find it, yeah, when I'm looking at a new platform, I'm thinking, okay, well, how have they used AI in this? Because I did notice and, riverside, I've said it a few times in the podcast sort of in the last 12 months, 18 months, when I first started using it, there probably was some AI. I can't quite remember, but it's come on leaps and bounds in the last like year, 18 months, and probably doing a shout out here to get some sort of commission or something. But it's come on leaps and bounds and I'm really excited to see how it's going to be in sort of the next five, ten years and where we're at. It would be very, very interesting in a good way, I would imagine. So with regards to opportunities for enhancing music, you sort of explore the opportunities AI presents for doing that with music production. Can you talk a bit about some of the innovative ways artists can sort of leverage AI tools to elevate their compositions and arrangements? What?
Bobby Owsinski:
I found most interesting, as a player myself, was something like Scaler 2, where that's a very, very deep program but you don't have to go so deep in it that you can't use it. But what I found that I really liked was the suggestions for chord changes that I never would have thought of, especially when it comes to turnarounds. It's like, oh, that is such a good idea and I think AI is brilliant for that, not necessarily for composition, and you don't want to take over your creativity, and that's a big fear that a lot of people have. But the fact of the matter is it can give you an idea where you thought, oh, I never would have went there before if it wasn't for that. Same with lyrics, you can take any number of lyric generators that are out there, and many of them are just based on chat TPD anyway. But if you're hopeless at writing lyrics, then it will write some for you. But if, in fact, you're any good at all, there are times when you just need a word or you just need a phrase and it will pop up something that you'll go. I can use this. This is great. So just in terms of how it can help you going to a place that you never thought of, but a good place. I think it's invaluable for that. But you have to approach it with that in mind, not like this is going to do everything for me, because, although it can, it's not going to do it as well as you.
Bobby Owsinski:
There's a big differentiation between and let me just go back for a second One of the things that people are afraid of, and I know. Before I wrote this book, I put a survey out to my list, which is quite large, and I asked them about this. I asked them what problems do you have with AI, or potential problems or fears? And two came up. One was it's going to take away my creativity, and the second one was it's moving so fast that whatever I learned today is going to be updated tomorrow.
Bobby Owsinski:
Well, in terms of losing the creativity, it's really good at doing 80% of the creative process, but it does require you to finish it off. Ai is good at certain things, but it's not good at nuances, and humans are really good at nuances. So that's a different differentiation right there. The second thing, the second fear about it moving so fast yeah, it really is, but the basics, like what I outlined in chapter one, for instance, the basics of AI. They're not going to change. So that's all the same stuff. We're building on that, but it's you know, there's a history there. That's certainly not going to change and the basics of AI are not going to change.
Bobby Owsinski:
So I think, if anything is going to get easier to use, the biggest hassle I have with AI in general right now especially with graphics platforms, video platforms, things like that is the interface is kind of clunky, and I think that's where the next leap is going to come, where we're going to find cleaner interfaces that will allow us to do more without having to, you know, basically put a chapter in a book as a prompt.
Marc Matthews:
Yes, yeah, I would agree with that. Some of the interfaces I've seen I know there's a particular one that is an image generator and I cannot remember the name of it, but I think you have to join the discord or something along those lines too, and there's just bits and pieces to it at the moment. That, to me, is a barrier, to use it. A barrier for me, that's mid journey, you're probably talking about yeah, yeah, I think it is yeah.
Bobby Owsinski:
And I find it clunky as well. It's like why should I have to do this? And then again, it's not easy, because when you generate something, you're generated generating it in the discord group, so everybody's seeing it. It's like I don't want people to see it right now. Yeah, I'm just playing with this, so experiment, yeah, yeah.
Marc Matthews:
Yeah, I can see how that would change with time and I think maybe at the moment, because it is such a, it's not really novel, like you say, it's been around for decades. But in that, in that sort of scenario, maybe it's a bit novel and they think actually at the moment we can, we can sort of have this funnel where we're putting people into this discord. But eventually, I think, as you say, there, with the interfaces getting cleaner, which they will inherently do, if you look at website web development, where, but way back when, you'd have to use something like Dreamweaver to put a website together, now you can go on Wix, godaddy, whatever it is, and you can just drag and drop and you can create a website in 20 minutes or use a template and have something up pretty quick. And also, with that as well, ai coming into web development as well. That's a different conversation itself.
Marc Matthews:
I love what you mentioned there about Scala 2, because I use Scala 2 and it's fantastic and I use it much.
Marc Matthews:
Like you mentioned there, I released an EP last year and I wanted to come up with a different base part for the second verse, albeit. I wanted it to be the same but with a subtle nuance change there, and I use Scala 2 for that exact reason. So I think it's fantastic for doing that, and I've often said to people as well if you're going to use something like that and you're going to use it to solve a problem and a really good way to look at it as well is to think, ok, well, maybe it solved it for me, let's reverse engineer it and see how it's come up with that, how it's got to that decision. You can use that as like a learning process, which I think is really, really cool. One question I did have for you and I've seen this particularly with social media is can you pick out? Say, if you were to let's think musically, like music in terms of a song, could you pick out something that's been AI generated at the moment? I think it's quite easy to do.
Bobby Owsinski:
Well, the easiest way is just by resolution.
Bobby Owsinski:
Because, AI cannot do high resolution. So that's the first thing. What you tend to get is very generic generic melodies, generic chord changes that don't change much. So you'll have one section and maybe a second section that you'll have, and they repeat over and over and over, and it's hard to get multiple sections that a typical song form would have. So, for instance, intro verse, pre-chorus, chorus, interlude, second verse and so on and so on, with an interlude, with a bridge, and then you don't get that with AI right now, so it kind of reveals itself right away. Now, that being said, there are a lot of songs that are kind of made that way, where they're just made on a loop.
Bobby Owsinski:
So, you look at it and you say, well, okay, it's not doing it, the traditional method that we used to do, and AI maybe is mimicking that. One of the things I noticed is most AI music generators tend to do electronic music very well. Okay, there's two reasons. One is it kind of lends itself. Ai kind of lends itself to those types of sounds. First of all, but second of all, a lot of the programmers are young and in that age that is their favorite type of music, so the platform tends to favor that somewhat. So what we find is many times we get much more advanced versions of songs that are two things actually both electronic and, surprisingly enough, some sort of hardcore punk, almost metal. Well, yeah, it likes AI is very good at chaos. Ai music is very good at chaos and distortion.
Marc Matthews:
Yeah, I can imagine that I haven't been in a metal band and the audience will have heard me say this many times chaos and distortion often comes with being in metal.
Marc Matthews:
So it's interesting you mentioned there about electronic music. Now, I suppose, in a way, because ultimately it boils down to ones and zeros, it's binary, isn't it the building blocks of computing and computer science? I suppose if you were to then, like I say, compare that with an acoustic led track or something along those lines, I suppose it's going to be harder for it to authentically replicate it. But I guess as it ingests more and more data, as the time goes on, it gets more and more sophisticated, then it will be interesting to see how it goes. What are your thoughts on sort of I've heard you mentioned this on the other podcast that I mentioned earlier about licensing vocals and using that as a creative tool, so being able to upload your what you perceive the vocal, the melody to be, then actually having say you wanted it's not going to be Elton John, but someone similar in that particular style. What are your thoughts on that?
Bobby Owsinski:
It's something that's being used now by professional songwriters, believe it or not, and the reason why is they want, if they're going to pitch it to, I'm going to say, taylor Swift, even though she usually doesn't take songs that she didn't write but if they want to pitch a song, they want it to sound like Taylor Swift, and so they'll get a vocalist that will sound like that. And that's generally the case where a pro songwriter is feels that they're going to have a better chance of getting a song placed if the particular artist can identify with it easier. So, as a result, getting a sound like vocal is very important to them. Rather than paying $100 an hour to a vocalist, it's easier to license a vocal that is similar and will do it in a flash. So that's something that's being used now, especially in songwriters, songwriter demos, things like that. But I see a big future in that, and already there are a number of platforms that are specifically designed to ingest virtual vocalists but also include them in on the royalty payments and also think about the image likeness portion of it as well. So it's being thought out.
Bobby Owsinski:
One of the biggest problems that they eye platforms is they don't understand. They always I don't want to make this blanket. Many times they don't understand the implications of copyright. So what will happen is the typical thing is you can generate this piece of music, but you can't publish it anywhere. Yeah, for free. Okay, you pay us $9 a month, and guess what? Now you can publish it on YouTube and pick another platform. Oh, you want to pay it. Oh, and, by the way, we own the copyright. Oh, you pay us $30 a month and you can own the copyright and you can do whatever you want with it.
Bobby Owsinski:
Now they're assuming that they could own the copyright, and I've talked to many top IP attorneys about this who say that assumption is wrong, because in the United States anyway, there's already been a ruling that you cannot have a 100% generated, ai generated song creation of any type.
Bobby Owsinski:
Really, that can't be copyrighted. So now here we have a platform that's doing that and by nature, it can't be copyrighted, but the platform is saying, oh, we own the copyright, so that becomes a problem, and we're seeing this happen in more and more platforms where you really have to dive down and read the terms of service to understand who owns what and what the implications are. Now, that being said, there are some platforms that are very upfront about it, saying you own everything, we own nothing, and those would be the ones I'd suggest to go to, because it's the least amount of hassle down the road. So I mean, nothing is a hassle in copyright. Until it starts to make money, there you go. If you're not making money, it's not a problem. As soon as you start to make money, then the knives come out.
Marc Matthews:
Yeah, and I read that throughout your book and it kind of goes back to what I was saying earlier as well about how the geographical, geographical location comes into play as well, because you can have something created on one side of the world and then you are located as the. I suppose the beauty of being connected via the internet is that I can then download it here, but I'm still under the copyright laws in Asia, for example. But, as you mentioned, then, until you actually start making like dollar on it, then you're probably going to be relatively okay. One thing I like about the idea of licensing vocals as well I'm not a vocalist. I wish I was, I wish I could sing, but I can't, unfortunately is that it's like another sort of platform for them to monetize their voice, because there's things like vocalizer, where I've used it, where you can go on there and then you could pitch your project and then they submit and then say okay and blah, blah. You go down the route of that. But now it's just another platform, I guess, for them to monetize their skill set.
Marc Matthews:
And I didn't know this until I read your book about the using artificial intelligence as well, with regards to backing vocals with Isotope Nectar 4. I didn't know that was a thing that you could do, and I think waves do it as well. There's a waves plugin yeah, I saw an advert on YouTube for it, but I think things like that are just amazing. I don't know what it sounds like at the moment. Have you used any of those tools?
Bobby Owsinski:
One of the things about Nectar that you have to understand is it's not giving you harmony, it's giving you background vocals. So it's giving you male, female and different octaves. Yeah, I mean, you'll have to repitch everything. It's not automatically coming up with harmonies for you. With the waves, one does the waves, one harmonizes things. I haven't played with that yet, but that's what I've read about it.
Marc Matthews:
Yeah, it looks very interesting. I didn't know that about Nectar 4. So I'm going to bear that in mind because I was going to dive into that tomorrow and have a look it's still worth having a look, because you don't always need harmony vocals.
Marc Matthews:
Yeah, very true. There is one big question. It's not my notes here, but this comes up a lot and this is to do with mastering AI and that has come on leaps and bounds. I know Logic just at the end of last year released its incarnation of a mastering Mixbus plugin, which is really good for, as a matter of fact, I put it at the end just to give it a reference of what it could sound like, and I've had that conversation a lot on the podcast about. Okay, well, you've got these platforms Cloudbounce, et al and you can use them and think, okay, this is what my master could sound like. But having read your book and you mentioned how they've come on a long way in the past sort of like five to 10 years compared to what they were before, where do you see that going? With regards to mastering, what is your advice if you've got an independent artist who is thinking of using AI and mastering?
Bobby Owsinski:
Well, if it's a major label project, they'll always go to a mastering engineer and they're willing to spend the money, so that's not so much a concern there. But if you're an indie artist, you're on a budget, then the online mastering programs, or even I'd say the online ones, are better than something like Ozone, because Ozone can take you down the rabbit hole if you're not careful and could actually make things worse instead of better. You just have to understand, to follow what it's telling you. But, that being, most people can't do that, unfortunately, they, you know they're going to tweak. But that aside, the online mastering platforms are really really good, and not only that, they're so close to real mastering engineer that the difference is almost like should I spend the money, the extra money?
Bobby Owsinski:
I've tested this myself with songs that I have that were mastered, and I would take the, the, the stereo mix, put it in various mastering platforms and compare it to the one that was mastered with an engineer, and it was really close, really close. The mastering engineer would beat it, but not by much. So you think, is it worth the extra money or is it worth the money? I don't have, you know, so that's always the question. That being said, there is a trick, and the trick is it needs a good reference. So the reference track would preferably be something that's already mastered and could be a hit. I mean, you can go to Spotify and download a track. Actually, it doesn't download the track, what it does is just copies the EQ and compression, so you don't have to worry about any copyright violation or anything there, and then it just matches it. That is a secret. So you find a song you think I really wanted to sound like that. You tell the mastering platform here's my reference, make it sound like this, and it does. So that will sound substantially better than if you just go to the mastering platform and say do it. So you know it doesn't take that much more in order to make that happen. So there are downfalls with the platforms. I think these will. I shouldn't say downfalls, it's more like there are some technical flaws that aren't thought through yet.
Bobby Owsinski:
One of the big things is mastering. Engineers are particularly good at putting an album together and there is some skill in this that is overlooked by almost everybody who isn't a mastering engineer Because they think, oh, it's just adding some EQ and compression and making it loud. No, a mastering engineer will take 10 songs and put them together and make them sound like they were all done in the same place and they have the same volume, they have the same tonal characteristics. This is especially important when you do have songs that are done in different places with different people. Even if it's the same artist, there's a different sound. The mastering engineer that's one of the, that's one of the expertise they have is to actually make sure everything sounds like it's in the same universe. So that's particularly good.
Bobby Owsinski:
And then if you're doing vinyl or CD and cassette, it's important that the spreads are taken into account. Spread is the time in between the songs, which producers have taken a lot of time figuring that out over the years. It would be a whole session just to figure out what the time would be. So the album will flow. So AI eventually will be able to do this, but it doesn't do it now and it doesn't do multiple songs, or most of them don't anyway. So that's something you have to consider. Oh, I'm doing this album. There's 10 songs I have to sound like you know all the same. So is my platform going to do that for me?
Marc Matthews:
Yeah, it's. We're going right back to what you said at the beginning of that there about the rabbit holes really interesting with eyes of type in particular, because I think I had this conversation with the mastering engineer around this time last year and it was very much the same, in that you've got all these options and it can almost be like, okay, well, they're all there and maybe I must use them, and then you could end up going down these rabbit and you get to the point of diminishing returns and then you could end up just destroying what it is you've actually done. So I think that's a key thing to be aware of and a really good thing point for the listeners as well, and also using a good reference, I think, like you mentioned there, just to hammer that home is key. But what you mentioned there about the online platforms like lander etc.
Marc Matthews:
Now I wonder from an independent artist perspective and I had this conversation, I think we I think it was with them, with Jay actually, on episode 130 about how the single and single releases now are incredibly popular. Obviously, we're still releasing EPs and albums, but singles seem to be the way to go, and I've seen that notably with independent artists, and I wonder if that's part of the reason why you see in these online platforms are getting more and more popular is because, okay, I'm just, I just need a single and to put a single out and it's, it's a lot quicker. It'd be interesting to know where, if there's a correlation between the amount of singles being released versus EPs and albums and online platforms. But I guess it's also to do with the fact that they are getting much better in terms of mastery.
Bobby Owsinski:
Yeah, yeah, I think that's true. One of the things that really confused me about some of the online platforms mastering platforms because they would have a monthly plan or a yearly plan where you can do thousands of tracks. Wow, and I would say to myself who does that many?
Bobby Owsinski:
But, then I had a conversation with someone at Lander who said oh, you'd be surprised. There are songwriters, for instance, that want all of their stuff and they're prolific and they'll be doing 10 songs a week, so you know, they need something like that. There's engineers, for instance, that will be doing mixes and they'll have five mixes of the same song that have variations, and they'll have to master them all in order to present them to the client and sign off. So there are many applications where you need that volume, although I don't think most humans need that. So those positions.
Marc Matthews:
Yeah, that is a lot. We're now 130 plus episodes in. I had this conversation probably about halfway into that number and it was with an artist and they were telling me how they had written. I think it was like a thousand songs in a year and I was just like, wow, that is something else. But when they explained it to me I was like, okay, well, I can kind of see how you've done that and I think maybe written was used very loosely in terms of like they'd started and they got the various incarnations of a release.
Marc Matthews:
So, yeah, I think that was used quite loosely, but it is quite impressive. I mean, it took me a year to do four songs. Yeah right, it's interesting. But also going back to what you mentioned there about how at the moment, obviously the subtleties of a mastering engineer being able to position the songs and having the space in between, etc I wonder if you'll get to a point whereby you could say, actually I like the song order of this album and the spaces in between, and then you feed that into an algorithm and then say I want my album to be similar to that. But I guess that's then when copyright might come in and you're having to present the album, but I wonder if that's an avenue it might go down.
Bobby Owsinski:
Well, it doesn't quite work like that, because what you're trying to do is, generally speaking, so there's a song at one tempo and you're counting it in the space one, two, three, four and then into the next song, so it's all based on tempo. Now, that should be easy for you know, an algorithm to figure out, for an AI platform to figure out, but it would mean that every song has a slightly different space in between. There are also times when you need crossfades, and right now we don't get that crossfades in between masters.
Marc Matthews:
Yeah, you know that and again.
Bobby Owsinski:
That shouldn't be a problem either. Down the road I just haven't gotten that far with it yet.
Marc Matthews:
Interesting stuff. Before we move on to the next question, there is another sort of plugin that I just wanted to get your thoughts on. I haven't used it yet, but there are plugins and I'm going to reference one in particular Goldfoss. I think I pronounced it correctly and I've seen that a lot on social media and producers, artists, engineers using it to finish their master. Maybe if you could just dive into a bit of the idea behind that and what maybe artists should be aware of if they're using a platform or a VST rather like that, something like Goldfoss.
Bobby Owsinski:
Yeah, there's actually another one that's similar. I just focused on it on my blog yesterday. It's called from Hornet, it's called Sleek, which is basically the same thing, and these are kind of a new kind of equalizer. They're tonal balancers. So instead of I mean we use brute force on equalizers. Where I'm going to boost 3k and throughout the whole song it's boosted at 3k, but it may not need to be boosted all the way through the song, I'd say, for a vocalist. Vocalist is moving on the microphone, so sometimes we get that 3k peak and other times it's perfect and other times there's not enough of it. So a tonal balancer uses AI to in real time figure out what the ideal tonal balance is, and it's across a lot of bands. I think with Goldfoss it's 128 bands that they're looking at and these are constantly tweaking as you go through to ideally balance it out.
Bobby Owsinski:
This may be the new EQ. Five years from now, we may not be using EQ as we know it today, because this is another way to do it. That's quite interesting and quite good. Now, obviously, there are still things that we use equalizers for. We use them for color. Sometimes we do need the big boost someplace or a big cut someplace, so it's very useful there, but for very small changes, and these are especially good at minute changes over time. These tonal balancers are very good. I think there's a focus right one as well. There's quite a few of them that have popped up in the last year.
Marc Matthews:
Yeah, it'd be amazing to see as again and what I mentioned earlier about, in five years time, to see where it's all at and in particular, with that as well, the tonal balance side of things.
Marc Matthews:
And once again, I think for the audience listening and I do this, and that's sort of an advocate of doing it. I think, if you're going to use that, I think in terms of like an educational perspective, is to break it down and think, okay, well, if it's doing that, why is it doing it? I think that's quite an important thing to do. And then, as I mentioned earlier about reverse engineering, it's much like if you're a programmer as well, because I've done a bit of that in the past you sort of like, okay, well, that script does that, that code does that, how does it do it? And then you can learn the subtleties and the nuances of it that way. But we are well aware of the time here. I don't want to obviously keep you too long to say, but maybe we could finish it off with some sort of myths and misconceptions. Maybe you could talk about some of the most prevalent misconceptions surrounding AI and music production that you've encountered.
Bobby Owsinski:
Yeah, the biggest one is that you could go to one of the more common AI music apps Bumi, for instance, ieva, one of those and just type in make me a Taylor Swift song, and it's going to come out sounding just like Taylor Swift, just in that simple prompt. It doesn't work that way and in fact, all of these sound likes that you hear are done the traditional way, where there's actually somebody that programs it up, that copies a particular Taylor Swift, drake there's a lot of them that are out there now. They copy a particular song or particular style and they program it up and then they get the vocal and they get one of these vocalizers. They get, you know, an AI vocalizer and and then they put Taylor Swift or Drake or Ed Sheeran or you name it, or John Lennon and they put it on top. So it's not like you're going to that particular AI and it's going to spit out something exactly like one of those artists.
Bobby Owsinski:
So that's the big one. It doesn't work quite the way everybody thinks. So we're once. Most pro musicians don't want that. It's mostly people that are not players that really want to do that.
Bobby Owsinski:
And that's what some of these platforms focus on. They focus on the non-player for the most part. So that's one. The other big one that we sort of touched on before was there's a resolution problem and if you think of it, if you're going to generate a picture, photo, a photorealistic image best way to put it it just has to do it once. If you're going to do a video, it's doing it 30 times a second, so that's not terribly taxing. But when it comes to audio and you really want it at the lowest possible resolution, that's acceptable it's at 44,100 times a second. So that puts a lot of stress on the AI and especially on the hardware involved. And then it's only 16-bit for the most part as well.
Bobby Owsinski:
Anybody that's been doing a lot of high-end recording, especially for record labels, knows that at least in the past 10 years and for me it's been 20 years I haven't recorded anything at less than 96k 9624. And generally a record label is going to demand that in their delivery specs. So that means that whatever you're getting out of AI, you have to up-sample it, and up-sampling isn't really giving you what you need. So that's a big limitation. We're starting to see bits and pieces of that changing in certain platforms, but it's going to cost you. So that's another big myth. That's there. Let's see what's the third one. Those are the two that come to mind in music generation anyway.
Marc Matthews:
Yeah, maybe there's only two.
Bobby Owsinski:
I think there's more. I haven't read that chapter in a while.
Marc Matthews:
No, what I was going to mention actually off the back of that there was and I think you've mentioned this before. You may have mentioned it in the book at one point on another podcast was the uses of it at the moment. I think one particular good use of it is for sounds in commercials and TV in the background.
Marc Matthews:
I think that's a really good. At the moment. That's one use of it I can see being really good. Now, if your profession, if your remit, is to create music, for that, I don't know how scared you would be of that, but I think that's where it could be used quite extensively. If you quickly want to get those kinds of West Coast customs, you get that hard guitar in the background. You can think, OK, well, AI, can you create 30 seconds snippet for me? I think that could be used quite a lot. In that I can see that going forward.
Bobby Owsinski:
Yeah, I think a lot of that. That's not a myth, that's actually happening. But that's one of the areas in music that is threatened by AI and that's what we'd call production music. Production music is music behind any kind of video, for the most part anything that's high-end. They're going to use real music, so that's not threatened there.
Bobby Owsinski:
But anything that's kind of low-end, especially like YouTube documentaries and things like that they're going to go for the lowest common denominator, cheapest thing they could get, and AI more than fills the bill in many cases.
Bobby Owsinski:
That said, that is not as easy to do as well, because if you have something very specific in mind and you go and you ask the AI generator to do it, it may take you a lot of tries to get in the ballpark. So therein lies a problem, and a lot of that has to do with the interface. The interface is looking for a prompt and for some Excuse me, I've got a bit of a cold here and for someone who isn't used to developing prompts and detailed prompts, it can take a while to really streamline something, to come up with exactly what you're looking for. I got to the point I got to tell you I ran into this. I wanted something for my podcast when I hit number 500, I was going to change the music and I did this. I went to the various AI platforms and I got to the point where it's like I can write something faster than this.
Bobby Owsinski:
I wound up actually taking something I did a long time ago and using that instead. It's not that I'm giving up on it, it's just like I ran out of time, yeah, so sometimes you think the AI is going to come up with something that's going to be perfect right off. That's maybe a big myth and fallacy right there that it's going to be fast and easy, and many times it's not what you think.
Marc Matthews:
It's going to take much more time. Going back to what you said about prompting, I've been having conversations for a while on social media about this and it kind of falls into how good a prompter you are. I think there is a key part in that, because I remember when I first started using chat GPT I can't remember I think it was like create a social media post for me and I pretty much just put that and off the back of that I was just like, well, this isn't very good. Then I realized actually there's quite a skill. It's quite a skill If you want to, because essentially it is. It's all down to you as a prompter. For all intents and purposes, they need that human input.
Marc Matthews:
How good you are at prompting is going to be key to what you get out of the end. In your book there's, in particular, with regards to I think there was I can't remember if it was a blog or for what it was for but the way you laid it out and saying, okay, you need to tell it who you are, tell it what it is and then describe exactly what it is you want off the back of it and then in bullet points as well At the end of it. I've never done this before with it is actually ask it, do you understand? I never thought of doing that, and just little things like that can really help. I don't know why I was going with this, but I think what I was trying to say is being the best. Being really really good at prompting, I think, is going to be key.
Marc Matthews:
I also think, going back to what you said earlier about how user interfaces and as soon as they get more sophisticated out of the wrong word, but less they're easier to navigate, more intuitive, and then it could have the prompts for you there on screen. I think it's going to be. It's really exciting to see how it could go, I think. With regards to interfaces, I think if they're made I'm going back to what we said earlier if they're made easier to navigate and then also prompt, I think you're going to get some really really good stuff out of it, really really good. One last thing I wanted to mention and you mentioned this in the book again is and I was incredibly impressed by this and I haven't listened to it yet is the fact that you can get a stereo Wi-Fi or similar and then separate it out into individual stems, which I think is amazing. I think it was done in the Beatles film, if I remember rightly, or something along those lines. Maybe you could talk a tiny bit about that.
Bobby Owsinski:
Separation is AI. Separation is very good right now and there's a lot of platforms that will do this and do it well. Ripx is one that everybody seems to like, but there's lots of them. Moises is another one. The process that you're talking about with the Beatles is the Peter Jackson proprietary process. Peter Jackson, the director from New Zealand.
Bobby Owsinski:
I went to a playback at EMI of the Beatles revolver what was done in Atmos and Giles Martin, the producer, was there and he was talking us through everything and he stopped and he said okay. So you have to keep in mind this is all a four track and they managed to spread things out, but very tastefully, I have to say. And then he said, okay, on track one it's two guitars, bass and drums all on one track, but we could separate all of those out, including on the drums, separate kick and separate snare. And then he went and he played them for us and it sounded so isolated it was amazing. So Peter Jackson's process is leaps and bounds ahead of everybody else, but there are lots of good ones and certainly you know if there's a couple I won't say faults, but a couple downfalls here, if you're using it, usually it will default.
Bobby Owsinski:
Almost all of them default to a certain number of stems, and that will be vocals, bass, and it might be keyboards or guitar and then other it will say, and that's kind of like everything else and drums, so you get five, and people want more stems than that.
Bobby Owsinski:
So I've talked to several of these companies that are saying, yeah, we're working on that or it's available, but it's only available in our private label or something like that. The other thing that people really want and it's funny because I was just talking to a sound mixer the other night for movies Alan Myerson, who's one of the top guys, does all the tentpole movies and he was saying something that I hear this from everybody what I want is I want to take background vocals or a choir and I want to separate those out, and none of them are good at that yet Now Peter Jackson's maybe. I haven't heard it being used in that case, but the separation technology is fantastic, except for something like that. So there are these things here that people want that it's still not providing, but for the most part I think it's fantastic. You know what it can do just about anything that's worth playing with, for sure.
Marc Matthews:
Most definitely and I like what you said there and I think it's concurrent throughout the chat today. It's all about yet, isn't it? You say it, you can't do it, but yet it's that growth mindset with AI and eventually it will get to that point. And I think being able to separate stems out like that, and I get this frequently. If an artist were to send me a wildfire and they say I've lost the project, I've lost the session, which happens so much. And I listened to it and I was like, yeah, you can do this, this and this. And I feed back and they're like, yeah, I don't have that anymore. So that's what it is, and in instances like that, I think it's going to be amazing. But it's really, really exciting stuff.
Marc Matthews:
And the book, again, just as I said at the beginning, it's absolutely fantastic. So I'm going to put a link to that in the episode show notes for the audience to go and check out. Well, bobby, it's been fantastic chatting with you today. I've been super excited about this and I think the audience is going to get loads out of it. Just for the audience listening, if they want to learn more about yourself, your blog, your podcast, where should they go online?
Bobby Owsinski:
If you go to bobyosenskycom, bobyosenskycom, and from there you can find the podcast, you can find excerpts from my books, you can buy the books if you want. You could find my podcast blogs, all that stuff, just from that one URL.
Marc Matthews:
Lovely stuff. I will put that link in the episode notes so audience do go check that out. And, as I say, fantastic resource and wealth of information there. So do go check it out. Bobby, it's been an absolute pleasure. I'm well aware of the time zone, so I'm going to leave you now to get on with the rest of your day and I'm going to have some dinner. So it's been an absolute pleasure talking to you and I will catch up with you soon. Thanks, mark.