Generative AI Ain't Got Nothing on This

0:00 Music Appreciation. What's good?It is Scoobert Doobert, and welcome back to Love Music More.What a guest Alex Delicata is.I can't wait to talk to him more.Hopefully I can write with him someday.Really, really impressed.

0:25 And it was great to pick his brain a little bit because I've had questions as a songwriterthat I finally got to ask in that pod and I hope that you enjoyed it.And it was funny because I love reacting to these podcasts because those interviews,I don't want to control the situation.I want them to talk.And then afterwards, I can digest it for that week or whatever and kind of like soak into

0:48 my brain and say, okay, what were the things that I'm stepping away from that conversationand I want to react to or I want to add additional questions to or it can play intothe storylines that I've been playing with in the previous solo pods.And that's what I'm trying to do here is I'm thinking about the role of the writer andall of the different things that we talked about from endurance athletics to taking walks

1:12 and parks to working with different people and creating positive environments for collaborationand not being precious with your art and allowing another team to take that and takeit to a different place that you never would have intended.All of those different through lines, I think that it plays back into that AI conversationin a really interesting way.

1:31 And I don't want to belabor AI because it's important, but it's not the only importantthing.So hopefully this is going to be one of the culminating episodes of this conversationat least for now.I'd love to get some more guests on to talk about AI and their perspective.But hopefully, this will kind of put a bow on that little cycle.

1:51 So I was thinking about, what if you took my head and put it on a completely different body?But I still be able to play guitar, or even better yet, pick somebody better at theirinstrument, John Coltrane.You take Coltrane's head, just like a Futurama where they have the brains in the jar.Just take that brain in the jar and put it onto another person's body, somebody that'snever played saxophone before.

2:22 What is Coltrane without his body, his hands, his mouth, his tongue, all of these thingsthat if you don't know saxophone, a lot of this stuff is going on in his mouth.Like saxophone, one of the cool things about it, and I really want to play it someday.I can't.I've tried to practice it a few times in my apartment, and I am horribly embarrassed because

2:43 I am really not very good at it, and I live on top of a bunch of other people.And it is just a very hard instrument to be bad at.I have it.It's staring right at me.I want to play it deeply, but another day, I digress.The limited amount of experience that I have on saxophone reveals how it's kind oflike taking your vocal cords and putting them up to your lips.

3:07 When you're playing a saxophone, you have that reed, and it's slapping against your lowerlip, and that reed is vibrating, and if you overblow it, the reed stops vibrating properly.That's when you get that squawk sound, that awful, awful intro saxophone player.I'm good enough to avoid that most of the time, unless I'm trying to do really high notes.

3:27 But there's a very low bar, because it's like the worst sounding thing.Really what you're doing is you're kind of like tearing your vocal cords kind of, youknow, it's your reed, but it's like the same kind of idea.You're overblowing, and you're losing control of that air.But a lot of the stuff that's going on with that is a very complex system.You're diaphragm, you're lungs.

3:50 All of these things are being controlled by your brain, but you're not really thinkingabout it.Coltrane, there's so many, like, neural pathways and muscle memories going on, if youjust took his head and put it on somebody else's body, I almost feel like you'd be missingmost of it.So like, yes, I have studied a lot of music theory.

4:15 Yes, Coltrane has studied more music theory, but that theory is only to inform our practice.It helps us figure out the next path.It helps us when we get stuck, when we lose our intuition to understand like whereto go.Or it helps us to edit when you have a strike, you know, of inspiration, and you writedown a thing and you go, what was that?You can analyze it.

4:42 You can understand it and you go, oh, this would be a cool way to mix it up in the B section,where this would be an interesting way to solo over this chord.That's where the theory comes in.But it's not a constant process.You're like, you asked somebody that's soloing here, are you thinking about the theoryas you're going?It's like maybe occasionally, if I get stuck, or if I'm in a jam session, I'm getting

5:08 hit with something new, a kind of dip into a bag, but really what I'm thinking aboutis making cool music.And that is like the crutch.If I'm thinking about that too much, I can't play.It's like, are you going to ask Steph Curry or Kobe Bryant, like what they're doingwith their feet?It's like they practice footwork like crazy.Yes, they know what they're doing, but in the moment, game seven of the NBA finals,

5:34 are they thinking about what their feet are doing or are their feet just moving?It's like, imagine a dancer who's like learning their steps until they internalize it.You take their head and you put it on somebody else's body, can they dance?Over time, yeah, they could build that back up, but what they're really building up is theirmind body connection.

6:00 So how do I explain to like a mixing robot, make my eyes get wide in this section?How does the AI grasp that?Because all an AI is, is it's trained.So you have an input, you have an output, you optimize the path in between for an output.That's all that's going on.But it only knows what you've trained it to do.

6:28 And yes, it can train itself in some of these advanced models and stuff like that.But ultimately, you're saying, is this a cat, yes or no?It says, this is a cat.No, that's a dog.It goes back and it goes, okay, is this a cat and it's like, oh, yeah, that is a catgreat.So now it has this pathway of like these things that look like this, that's not a cat.These things that look like this is a cat.

6:48 And so then it can have within some realm of probability, ability to identify if it's acat or not.That's rudimentary AI, that's the kind of stuff that my friend made in undergrad.But how do I explain to an AI?Sometimes choruses shouldn't be big, they should be small.When does it know to do that?I don't even know when to do that until it hits.

7:18 Or how do I explain to an AI that sometimes a vocal should be out of tune, or that sometimesa note should deviate from the key.Even if it's not written that way in the musical score, that that scoop or that blue noteis the soul.You listen like blues music, there's some thing that's magical in there.And what it is is it's playing with pitch, it's not entirely diatonic.

7:45 And yes, you could train a robot on playing with pitch or not being entirely diatonic.But do they know when to employ that?And do they know how to create musical surprise?Sometimes there are two right decisions as well.There's a decision or a B decision, and you have to decide which one has more vibe, whichone gives you more mojo, which one makes you feel more.

8:12 So you have two different vocal takes, both of them are in tune, but one of them playswith the beat and the articulation slightly differently.It says its words, the lyrics, slightly differently.Maybe it's a little bit behind the beat, it slows down the rhythm, or maybe it slursthings together so it isn't quite together, you know, it makes things, or the opposite.

8:37 It's highly articulated and it's hitting every little thing.These nuances, that's what the producer ends up having to decide, you're going througha bunch of different takes.So the singer is saying it 20 times, two of them are great, which one's greater.I think a robot would choose the more correct version, the one that fits the model thatyou trained it on most.But is that the best?

9:04 And does that music change all the time?It doesn't preference change all the time?It's a fluid narrative.Like how about this?Background vocals, they're usually brighter, but should they be always?Do you always want every background vocal that you ever hear in the rest of your lifeto be bright?

9:28 Really high and tinny, every single background vocal, every genre, do you want that?Or even like every rap vocal sounds like this, every rock vocal sounds like this, isn'tit cool when rap vocals sometimes sound like rock vocals or if rock vocals sound likerap vocals?Why would we want to put ourselves in certain tributaries, doesn't that kind of like missthe whole point of where music is right now?

9:53 It's not really genres. You know, like a good example is sometimes I like to have things change mid-syllable.It's like you'll have a background vocal that's super, super reverbie and then it growsdry right in your face.Because it makes you feel something, does a robot know to do that?That's a chaotic decision.And how would they be able to tell that this thing makes the listener feel moved?

10:23 And then said they go with what's more typical, what's most right?Again, we learn these rules so we can forget them.We fall back on them when we need to go quickly or when we need to do something when intuitionfails us.Intuition disappears.It's like writer's block.But we don't learn these rules as a crutch.It's like some of the music that I made and other my friends made in music school where

10:50 learning all the rules sounds like crap because it sounds like rules.And it's like, that's not hating on the rules.I love that stuff.I still read it.I've got a counterpoint book on my bedside right now because I like it.It's fun.It's interesting.But you learn them so that you can internalize them and forget them.That's when you get good.It's that it's not you don't forget them entirely.

11:16 It's that you don't really think about it.You start incorporating elements of that or pushing off from it in the same way thatlike you look at people like Picasso or whatever, you know, people that deconstructed art,the best ones understood how to make the art that came before.You understand the two and three point perspective so that you can break it.

11:38 But you can also do it and the beauty is is when you can do both when you can push offagainst it and then lean right back in.That's when you get chills.Like, here's another example.There's something that I've set a ton in this pod.Control your highs.Control your lows for mixing.And that's the high end frequency and the low end frequencies.Control those.You mix on the extremes to start.

12:04 But I've been in situations where if you just across the board say snares don't havelow end or high hats don't have low end or floor toms need to be gated and cleaned up.So they're not always ringing and just bleeding out of the microphone because a floortime is usually hit like a few times in a song unless it's a certain type of floor tom feel.It's like a doom, doom, like that's the whole song.

12:25 But sometimes when you turn off that microphone, the drum sound sucks.Sometimes you need to leave that on even though it's wrong because there's so much vibecoming out of it.Like picking up something weird about the room or the character, it's adding a differentfrequency.But if we trained in AI, we would train it on its most likely scenario of floor tom, gate,snare drum, you know, high pass filter under 100 Hertz.

12:56 You might ruin the thing.You might lose all of the mojo.That's what the mixing engineers really get paid for.It's not just cleaning stuff.It's falling your gut, your intuition searching for vibe.And honestly, the cleaning the stuff, that's what the robot should be for.Or this is an engineer, but I've never had one of those, but I have been one of those.This is an engineer.That's basically what you do.

13:25 You clean, you color, you sort.I actually, I think I should talk to my buddy Riley.He's my mastering engineer for all my stuff, and he just started an AI company.But it's one that I agree with, where it sorts out different types of instruments and stems.And it can file name and file organize all of your stuff, which is really, really nice.

13:45 If anybody's done this kind of stuff before, sometimes you'll get a session, then it willbe like, audio one, audio one, one, audio one, one, two, audio one, one, two committed versiontwo.Like, oh my gosh, what are all of these?And this would convert it to like drums, bass, snare one, snare two, like that kind of stuffyou can actually use.

14:05 That's good AI, you know, enables me to like focus less on the crap and more on the creativity.Awesome.You know, and like put it another way, could an AI take a walk and get new perspective fromnature?Like we're talking about in that last spot.

14:29 There's a, there's a great Haruki Murakami, one of my favorite authors wrote a book calledWhat I Think About When I Think About Running.And I'm not necessarily recommending that book.I loved it because I actually listened to it as an audio book while I ran.And it's not necessarily for runners that just like run here and there, like get ina five mile or once a week or whatever and post it on Instagram.

14:57 It's for the people like I'm about to run 19 miles tomorrow and I'm scared as hell.It's going to hurt.I might injure myself.I would be pushed to my brink.That's what it's for.And that's where I grow as a musician.It puts that into perspective.

15:23 It's like running is such an analogous hobby because I hate it as much as I love it sometimes.Sometimes I hate it more than I love it.A passion equal or greater to my love.And I feel that with music.Anybody that's really, really given themselves to music, you felt that.

15:55 So that's why when I look at generative AI, I smell arrogance, like real arrogance.Some of the people that have thought of these things, have they bled for music?These programmers really know what it means how truly awful being a musician can be growingas a musician, how painful it can be, how embarrassing it is.What you go through to do this.And I know that it sounds like melodramatic, right?

16:34 But imagine giving up so many different future paths for something that no one cares about.Like really giving hours and hours and hours and hours of your life to this thing and feelingembarrassed, super embarrassed along the way.

16:58 Being put in these really tough situations where you play a gig, the biggest gig of yourlife.A showcase for a bunch of people.They don't even turn on the mains.And then you still have to pay the band $400, $400 that you don't have.Or when you get paralysis and you stop being able to play guitar and you lose all of yourgigs because nobody's loyal, so you have to go on Craigslist and find new gigs.

17:33 And then you're in gnarly dive bars playing four or five hour sets, loading in the PA justto make a hundred bucks, getting back home at four or five a.m. and going to school againat nine a.m.

17:47 repeating, doing that for 15 years, that's a little exaggerated, but I guess my entire music career, like since I've recorded it's been somewhere around 10, 15 years.It's a, if I could have, if I just would have taken a fraction of what I've put into musicand just applied it to computer programming, I'd be very rich, very successful.People would be demanding my skills.

18:17 Instead, I put it into this thing that so many people want, there are very few spots.It's very ephemeral.And there's a lot of crossover with athletes.You're gunning for these little tiny slivers of hope.But the beautiful thing about music is it's not really about money or success.I'm able to see colors that other people couldn't see.

18:46 I'm able to hear things that other people couldn't hear, I'm able to make stuff.Real stuff that is like, nobody else has done.Here it is. You can listen to it now.That's what frustrates me the most about generative AI.This is complex. That's an understatement.It's like looking at a brain, maybe even bigger than that.

19:14 You know, I talked to my neuroscience friends and they say the stuff that we hear in the mediaabout neuroscience and it's reading brains and whatever that's utter crap.We know very, very little in order to do that kind of stuff.It has to be in a very, very situational, highly trained, it's very stochastic and open.And there's all sorts of noise and all these things.It will get better.

19:37 Yes, but there is a agenda behind it all to try to get more capital.The people in charge of these businesses want to raise hype in order to raise capital.Totally understandable.Not even saying it's that bad.But the media also likes hype.But the people that end up investing in it get swept up in the hype but also believe thatthey're experts, these initial investors, these capital people.

20:09 But there's a real hubris going on with all of this generative AI.Over and over and over again.I'm reminded how small I am, how bad at music I am.But I wake up every single morning before I go to my job that I don't want to go toto pour another grain of sand onto a massive beach.And then I get back home and I pour more sand onto that beach.And then I'm here on my weekend pouring more sand onto that beach.

20:48 And someday I want to build a tiny sand castle.And maybe it'll get big enough that other people might admire itbefore it swept away back into the ocean again.That's what I'm doing.It doesn't matter how many songs, it doesn't matter even the reach.It matters doing that act, giving to music itself, repeating, showing humilityand growing every single day of my entire life.

21:21 Be it with an instrument in my hand or walking around in nature.And to say that you've created generative AI that can match that, I call BS.There's no computer that will go through that level of pain,the musicians, like myself and others.Go through every single day to bring this art form to you,to share it with you because it's ours and then we're sharing it.

21:53 Sometimes I get tempted to never release my favorite songs.I've seen other musicians do this.That's like a crime.So hopefully this pod helps you love music more.You can tell hopefully that I'm very passionate about this.I love that conversation with Alex.And I hope that I can do my part to explain where all of us are coming fromas creators and hopefully make something worth listening to.

22:31 I'm going to keep trying until I die.And hopefully that's a long, long time.And hopefully I can make a ton of music.Because eventually, like Alex was saying one in 30, it's like all right.To at least 30 a year, one of them should be good.Hopefully 30 of them will be.I'm going to try my best, try to make the best thing I possibly can.It's not going to be phoned in.It's not going to be like this AI stuff.

22:59 I'm not going to be training on models.I'm going to be training on vibe and on what's out in there in the ether.What's out there.You know, in art itself.We're all kind of connected to.If you'd like to check out my music, you can check it out at scoobertdoobert.pizza.You know, like and subscribe and all that stuff.So people can hear this weird podcast.

23:28 Half the time, it won't be that weird because I'll be talking to people.Other half I'll be all in my feelings like this one.I hope you enjoyed it.You know, feel free to reach out.I'm happy to have these conversations.Obviously, I love talking about this stuff.But yeah, try and sleep on it.Coltrane.Put his head in another body.Does he play in the same way?Lose the calloused lips and the calloused hands

23:56 in the thousands and tens of thousands of hours ofhitting those pads.You just take his head.Is there really a mind-body split?Could he really be a cyborg?And still be yourself.Huh, think about it.Check out my music scoobertdoobert.pizza.See you guys next week.Bye.