Dr. Bill Evans on Using AI for Good Not Evil – Insights In Sound 221 – S23 E1

0:00
Welcome to Insights and Sound, where we talk to the people behind the scenes, behind the technology, and behind the
0:06
music. People you may not know, but you should. [music] And please check out
0:11
insightsandsound.com for a full list of our podcasts and YouTube series. Insights and Sound is made possible with
0:18
the support of our friends. I don’t know how this heard
0:26
than a thermometer. Uh but [laughter]
0:31
He is a multiaceted and multi multi-dereeed um highly educated student of both music
0:40
and human psychology. And uh very very interestingly enough uh
0:47
one of your creations which most people have heard of but most people don’t know anything about or know very little about
0:53
is Prism. And uh we’re going to get into that in a moment. Prism actually stands for I I
1:02
looked this up uh performance restoration and isolated structural modeling
1:08
which is quite a mouthful. And so so let’s just start off with a
1:14
little bit of a um a background on you and how you came to uh how you came to
1:22
have created this this thing.
1:38
Well, I had worked on two uh PhD projects. One
1:45
was at uh University of Glasgow and that was creating uh the virtual audio
1:51
workstation. And that was this idea that instead of
1:57
having to directly affect data, so if you wanted to change the dynamics of a
2:03
signal, you actually had to turn knobs on a compressor. Instead, we would virtualize the whole uh DAW
2:12
and have the parameters just that are useful that you want to use. And part of this was to unite audio
2:20
tracks and MIDI tracks to a single track type. And it was a a virtual audio was
2:29
displayed uh in a 3D representation as clay.
2:35
And you could on this virtual surface you could feel it and and touch it. It could project um a force field in into
2:42
space uh called the volutric haptic display. and you could shape and mold the sounds
2:49
and and move notes around. Uh that sort of thing.
2:54
Uh and then I went to MMU and my plans for what I was going to do
3:02
there uh got derailed because of something that happened with my band. So
3:07
I I started a band uh progressive rock super group called a bunch of pack musicians. All of them. Yes. with um
3:14
Mike Portoi and Steve Morse and Neil Morris. Uh
3:20
probably not easy to keep them under control either. [laughter]
3:25
Absolutely. Um yeah, pretty much it was not easy to get them,
3:30
especially because I’d never done [laughter]
3:35
Well, you kind of you just kind of wind them up and and you just got to let them go. Uh but the and we had uh uh Peter
3:43
Collins as the first um producer on the first album and that album became
3:52
uh against all possible odds really successful. Was that the Live in Europe one? Like you could buy I could buy it in
3:58
like Walmart and Best Buy and there was like a section for us even though there was there was like one little rack of
4:04
CDs but like somehow we were in there. Yeah, no kidding. And
4:09
uh no that well that that one went to number one which is bizarre for a
4:15
progressive rock album. Uh but uh so you
4:21
know we were humming along and we did this uh live performance at the Z7 in
4:28
Switzerland and it was a big production and it’s great show and afterwards
4:36
the audio company said well we have a problem that’s not a small cameras and I was
4:42
like oh okay three of the four main cameras didn’t record anything.
4:51
Yeah. Okay. [laughter] Well, you see it’s like this. Uh, and I was like, “All right, well, at
4:58
least we got the audio.” And Bernard is like, “About that.”
5:07
[laughter] And
5:12
a huge chunks of the uh of the audio recording were either garbled or just
5:21
noise or what sounded like noise. And I
5:28
will typically with the with the band’s music, I’ll start with a a production process
5:34
and but we all realize [laughter] there’s like we can use this but we’d already spent the record company’s
5:41
advance. So they were like, “Here you go, Bill.
5:47
Sure. [laughter] Figure it out.” Uh, [clears throat] all of us uh didn’t want to do a
5:55
re-recording and like pass it off as a live album.
6:01
So I stopped the research that I was doing and tried to see is there a way to
6:10
recover performances when they’re so garbled that you can’t really
6:16
hear what’s going on or you can’t hear anything.
6:22
And that rabbit hole uh ended with performance restoration which was my
6:27
[laughter] uh PhD thesis at MMU uh which was entirely done out of
6:36
uh terror not having to deal with this uh because mascot label group is very
6:41
generous. Uh so the the idea was can we
6:46
use science to do this? And
6:52
one of the things that we had to do was predict what one the these people had
7:00
played in spots where there was just nothing. And I did use AI with this, but the AI
7:09
didn’t actually generate the audio because it had to be exactly what they played, not something plausible that
7:16
they might have played. And there’s a lot of things that happen in
7:22
our brains when we’re playing music that leave a trace
7:28
that allow you to reconstruct what came before and what comes after.
7:35
So when you play music there are two cognitive models that form in your head. One is a tree of the motor
7:43
movements. At the end of the tree are your fingers and then your wrist and then your arm and then at the top of the
7:50
tree is your shoulders. And at the same time there is a a tree that’s sort of
7:55
like shankarian analysis. It’s got notes, then motifs, themes, etc., etc.,
8:02
all the way up to uh, you know, uh, choruses, etc., all the way up. And when you perform music, you synchronize both
8:09
of these and then you download the motor movements into your fingers.
8:19
When those things get out of sync is when you make mistakes. But all of that follows a very if you
8:27
can find it a very specific mathematical pattern. So some so you might not be able to
8:33
figure out the notes but you might be able to figure out the finger movements.
8:41
And so there was some machine learning there in terms of being able to figure those things out um and match them.
8:50
And with the help of my assistant Tom, uh we were able to reconstruct the entire
8:57
performance without any overdubs. And a lot of the garbled stuff,
9:05
you’re able to find some things that are correct. Even when a musician makes a
9:10
mistake or in this case the audio just isn’t there, the timing might be correct
9:17
or the timing is not correct but the note is correct or the note isn’t
9:22
correct but you can tell what string that they were playing. And so if you can identify one of these
9:29
variables that’ll give you a foothold to identify more of them. And then you mix in the
9:35
cognitive modeling of what came before and what came after and
9:42
you can arrive pretty closely at what exactly was played.
9:47
And it had to be because these guys are not easygoing about this stuff. It has to be
9:55
exactly what they played because you know they it’s a matter of integrity for them unfortunately for me. Uh, but the
10:03
most extreme moment was when there were seven seconds that were just completely
10:09
missing from one of Steve Morris’s guitar solos.
10:14
And Tom and I were able to compute
10:21
exactly what he would have played, right down to like how he changed the angle of his pick and then to render the audio
10:29
that corresponded to that. And I sent like the whole 30 secondond
10:34
solo to Steve and he came back and he said, “I don’t know what 7 seconds that
10:42
was because that whole solo is what I would have played.”
10:49
And this was the first time that I realized that okay maybe this is actually going to work
10:55
that it was possible to not generate someone’s performance like
11:03
traditional gen AI but to use AI to help
11:08
figure out what someone would have played and then use more procedural and
11:13
mathematical things to actually create the audio because it has to be deterministic. It’s not deterministic.
11:19
It’s not what they played. It’s something they might have played.
11:25
Okay. This brings up a whole lot of questions. Um, obviously I mean I’m I’m
11:30
I’m waiting into a hornets’s nest here in many ways, but um from your
11:36
description of this uh it seems to be sort of
11:42
it’s it’s neither predictive nor generative AI and yet it’s a little bit
11:47
of aspects of both. Um, but of course the question that almost anyone,
11:54
especially the more lay person who is only peripherilally familiar with AI, is
12:01
going to ask is how do you know? How can you speculate about something that may
12:07
or may not have been? Um, you know, and and perfect example, I mean, you know,
12:14
you can talk about someone like Steve Morse who is a very disciplined and um
12:21
very detailed player. I mean, you know, his his riffs are not sloppy by any
12:26
means. He’s uh he is precision personified in a lot of ways, but you
12:32
know, could you I guess, you know, my question is could you apply the same thing to a Miles Davis? Could you apply
12:39
the same thing to someone who is um very very outside and less disciplined and
12:46
less predictable in that sense? That’s a really interesting question.
12:51
One of the things that I found when I was doing my uh cognitive research was
12:59
don’t remember if it was it was m either Miles Davis or a jazz musician of
13:05
similar renown for thinking outside the box, right? I’m thinking like a coal train or
13:11
you know any any of those kind of cats, you know. Yeah. Uh and when I did the analysis
13:18
and AI is really good for this of just saying here are the things you should look at not drawing the conclusions but
13:25
saying look at the right and everything they played was some
13:31
combination of just 45 different riffs
13:36
[laughter] which blew my mind. I don’t know how Miles would have felt about hearing
13:42
that, but okay. It’s it’s math, you know. You can’t argue. That’s the thing about math is is
13:48
it it it exists whether and I had a hard time believing that,
13:53
but that ended up being a really key piece of research because it meant that
13:59
even the most inventive, creative musicians had patterns and you could
14:05
find the patterns. Sure. And music is mathematical. We can’t argue that. I mean that’s that’s
14:11
been well established and and you know that’s why so many musicians are you know sort of torn between left and right
14:18
brain. Yeah. You know but but I still I I wonder
14:23
about a lot of this in terms of and again you know I’m I’m I’m picking
14:29
this apart and I’m playing devil’s advocate obviously because that’s part of what I do uh being the p the being
14:36
the pedantic person that I am. But um I I’m just wondering about this because it
14:42
seems almost too it seems almost too cut and dry for
14:47
something that is or can be as as unpredictable as music. I mean in a
14:55
sense, you know, I think aren’t you kind of pissing off musicians by saying, “No, I I can apply simple math to almost
15:02
everything you do.” you know, because we like to think of ourselves as free
15:07
spirits, you know, and unpredictable. Well, I think it
15:15
Well, the first thing I can say is is that they’re they’re usually really happy about it because I’ve been called
15:20
in to restore something and so I, you know, I’ve worked with like Steve I and
15:26
John Puchi and Albert Lee and Jay Graden. Oh. Oh. From a practical perspective, I
15:32
think you’re I I think this is wonderful. I’m just wondering about, you know, the the idea that we can that
15:38
people would question the the authenticity in that sense, you know, how do you know?
15:43
Absolutely. Until the artist actually hears it and then the artist is like,
15:49
“Oh, okay. That actually Yeah.
15:54
Now, okay, but is the artist saying, “Yeah, that’s what I played.” or are
16:00
they saying that sounds like something I would have played? I guess that’s the question.
16:05
I think with with all of these guys that it has to be that’s what I would have
16:12
played exactly or it’s, you know, it doesn’t get
16:17
released. I have never heard with any of the people that I’ve worked with someone say
16:23
that’s good enough. Sure. Sure. and and maybe it really does
16:30
get to a point where it’s really kind of splitting the difference of, you know, because the truth of the matter is how
16:36
do I know what I would have played? So, there’s that argument, too, is there not?
16:43
Yes. And and if we have time, I can show you some examples of that. But
16:49
everything I do isn’t audio restoration because I was able to take what I
16:55
learned from that and start answering some other questions. Like
17:01
my whole I mean my whole thing is performances. That’s what I sort of have devoted my
17:08
career is in how do we emphasize and pull out genuine performances.
17:14
Mhm. [clears throat] When people record an album and they’re not in the same room,
17:20
you can tell. Oh, definitely. Definitely. But because
17:26
people don’t music isn’t monetized the way that it used to be,
17:32
no one can afford to do that. And there’s also this
17:37
expectation of perfection. So even if musicians could get together, they don’t want to because they’re going
17:43
to read about it on Reddit, you know. [laughter]
17:48
And so Oh, it’s sad but true. Yeah. Mhm. I said, “Well, what if I use the same technology
17:56
to change the performances?” Now, this isn’t changing the notes that
18:01
they played. It’s more subtle than that, but if they had all recorded in the same
18:07
room, would that be possible? And I think the the first one
18:15
that I did was uh it was John Ferraro, Sterling Ball,
18:23
Steve Lucther, and
18:29
Jim Cox. Wow. And the question was, can we do that?
18:35
And they they came back and they said, yes. Um, and we That’s a tough room.
18:40
That is a very tough room. Uh, because the next one was Jay Braden.
18:47
Okay. [laughter] Uh but we we did a whole a whole album
18:55
um and uh it’s uh Sterling uh it’s called the Mutual Appreciation Society
19:01
and uh it’s a wonderful album if you really like uh guitars and and
19:06
incredible performances. Uh but that was an example of a different application of the technology
19:13
is how do we put people in the same room?
19:19
Yeah, I mean it’s interesting. I um I have been working with a a camera company who are partnering right now
19:26
with a virtual a virtual reality organization where they’re taking uh
19:33
literally dozens of these PTZ cameras and they’re forming sort of a cage where
19:40
they can then scan every part of the body and create the illusion of people
19:45
being there in the room together. And they’re doing that with immersive audio. So, uh, this is right up that
19:52
same up that same alley. And I’m always fascinated by that because there’s there’s, you know, obviously we joked
19:59
for years and years about phoning in our parts. But the the challenge there really is recreating not just the
20:08
playing together concept, but the energy, the the mental energy of playing
20:13
in the same room. That’s the stuff that’s lacking. I did um a demo of a
20:19
system that I developed in 2008 which was supposed to address that. And
20:26
what it did was it created live realtime representations of the
20:32
musicians that were playing all over the world in the same room
20:39
holographically or no. Um, I used uh data projectors
20:46
and measured the cameras so that the uh final images were exactly life size.
20:55
Uhhuh. And then the the British military gave me this very strange substance because I
21:02
needed to have cameras behind the screens, but the screens needed to be
21:08
opaque. And so they lent me this weird substance that was completely transparent on one side and completely
21:14
reflective on the other. Not like a mirror because a mirror one
21:20
way mirror makes it darker. Uhhuh. Um and we did a um the first performance
21:26
of a song by Carrie Livrren uh which was his uh his sequel to Dust in the Wind.
21:32
It’s a it’s a beautiful song called The Navigator. Um. Uhhuh. And I was very interested in what is it
21:40
that makes us feel like we’re in the same room.
21:45
Well, the big thing is eye contact, which that would then address. That’s right. I had to put the cameras
21:51
right behind people’s eyes, right? Because that’s one of the things that, you know, is so infuriating and uh
21:58
and frustrating with any kind of a Zoom meeting or anything like that is the idea that there’s this illusion of eye
22:05
contact, but you know, I’ve used a lot of the the different technologies like
22:11
session wire and stuff like that, and it’s a it’s a wonderful thing that you feel like you’re in the same room at the
22:17
same time, but that eye contact component still hasn’t really been addressed to the degree that it could be
22:24
and should be. I I agree. Uh
22:29
I I ended up building these these sort of virtual rooms after that. And in
22:35
fact, we’re in the one that’s what this data projector is for um is to create
22:42
um these virtual rooms. I’m really I’m really I’d love to talk to you some more about that. What I discovered was that
22:48
there was aside from eye contact, there were a number of other cognitive switches
22:54
and that where people I think had made it too difficult
23:01
was that they were trying to actually create the people in the room and that’s a hard problem.
23:08
But if instead you focus on throwing the cognitive switches and working backwards from there,
23:14
that works. Ah okay. Uh so uh yeah that that’s that’s uh I’d love
23:22
to talk to you more about that. Uh something yeah I mean to me that is you know that
23:27
has always been I mean you know both for myself as a musician and as a producer having worked with a number of other
23:34
artists and whatnot. I think, you know, one of the one of the big causes of
23:40
chagrin for me is the lack of playing together. And the big thing that we lose
23:47
is that spontaneity. It’s not just the spontaneity of the musicians playing together. It it’s the spontaneity of
23:54
their reactions to each other, the chemical reactions in the brain. the whole idea that, you know, the drummer
24:00
does something and and I, as the bass player will will riff along with them the next time they go through it or
24:07
something like that, you know, it’s it’s this it’s something that’s very ethereal and
24:12
very hard to really delineate what it is, but it’s something that just does
24:17
not really translate well to working remotely.
24:23
And that to me would be probably one of the biggest challenges to address in something like this. I could send you one of these systems
24:29
and it works. You will feel like you were cuz again I worked backwards from the the the the
24:36
cognitive um one of the things I found was doesn’t need to be 3D.
24:42
I could see that it doesn’t, you know, the thing is I could see having, as I
24:48
mentioned, you know, I’m I’m a I’m a big fan of uh session wire, for example, because it is it is that it is one of
24:55
those big steps in terms of allowing people to interact in what appears to be
25:01
real time. And you know, I I think that what was more important to me than
25:08
having a 3D I don’t need a hologram of somebody in front of me. What I need is just the feeling that I’m there with
25:14
them in the room. And that doesn’t necessarily mean that I need to see them and pretend they’re in the room. What it
25:22
means is that I need to have those mental components that we’re talking about, whatever they are, however you’ve
25:28
identified them, that I need to have that that that vibe.
25:33
Yeah. Essentially, and whatever it is that creates that.
25:39
But an example I I like to use is one with 3D vision because every there’s
25:45
been a lot of work of people trying to create free floating 3D displays
25:50
and everyone says well you know it has to be a different different image in each eye
25:57
but no it doesn’t. If you were if you take one hand and put it in front of your eye
26:04
does the room look any less three-dimensional? No, it does not.
26:11
And you know that that’s the kind of thinking I think you have to use to solve the problem you’re talking about
26:17
because in reality physics says you can’t do that.
26:22
But you know we’re we’re just you know meat with electrical signals and chemicals
26:30
and that’s depressing. I know right. [laughter]
26:35
I got I got to I got to tell you a quick Mike Port story. Um because you were saying about you
26:41
know when people are are not working uh in the same room and when we were recording the first album
26:49
um everyone really everyone was there but Mike said I’m going to go do you know just record my drum parts myself.
26:56
And Peter Collins and I were like well you know we really want to have that interaction. And Mike’s like that’s
27:02
okay. And I asked Mike, “Can I be in the room
27:08
with you when you record?” You know, cuz like secretly I’m like this big fanboy, you know, and I’m like, “May I?” And
27:15
he’s like, “Well, I really don’t allow that, but okay.”
27:21
And I’m like, [clears throat] “I’m going to get to see how Mike structures and composes these incredibly musically
27:29
perfect parts, you know?” So, I get like a a a a
27:34
sandwich and and a beverage and I’m going to sit there while he does the first song.
27:40
So, Jerry Jerry rolls the tape. Now, the band had just written the song
27:48
and made like a rough demo and that demo hadn’t even been assembled yet. So, this
27:54
was the first time Mike was actually hearing the whole song put together.
27:59
So he’s still sussing out the part. Well, hold his beer. So [laughter]
28:08
Jerry crosses play and Mike is hearing it for the first time
28:13
and he records an absolutely perfect drum part.
28:19
And then he says next and we go to the next song. I left after the third song because it
28:26
was boring because Mike there was no process that you could see. He would
28:32
hear the song for the first time and as he was hearing it he would play
28:38
these incredible and that was really sort of my introduction to the level of
28:45
just how much better these guys were than I knew that it was even possible to
28:50
do. Because the other part of that is that when you hear the album, you hear
28:58
like Mike and Dave and Steve playing off each other. Yeah.
29:04
But they never recorded at the same time because Mike also knew exactly what they
29:10
would play even though they hadn’t played their final parts and played the responses.
29:17
Do you think that’s just a matter of him being very familiar with them having played with them for so many years or is
29:24
this more on him as a predictive musician? I mean, why why would
29:29
something like that even happen? I think there’s a picture somewhere in an attic of Mike getting clumsier. He
29:36
he’s he’s just that they all are. They’re just that good. And I as a
29:42
scientist, there was a point at which I couldn’t explain it. anymore,
29:48
but I assure you that [laughter] they haven’t.
29:54
Okay. I mean, you know, again, there are a lot of people who probably are not familiar with these particular
29:59
musicians. I will say that I’ve been, you know, I’ve been a a Steve Morris fan
30:04
since since the 70s. So, um you know, I
30:09
know these guys. I know I know the type of players they are. Um, would this be a
30:17
similar situation with completely different types of musicians? These guys are, I don’t want to call them fusion,
30:25
but they have elements of fusion. They have elements of a lot of jazz. Uh, they are technically
30:32
in the top 1 or 2% of musicians in terms
30:37
of what they can execute. And that does impact the type of music they create. Would this work with more,
30:46
shall I say, typical um meat and potatoes kind of musicians?
30:51
Would it be as applicable? Would what be as as applicable? Predictive analysis.
30:57
Oh, absolutely. It’s easier uh with, you know, the more close you
31:03
get to how poorly I play, the easier it gets because it’s more
31:10
sort of instinct and less finesse in terms of of what you’re doing. You’re
31:18
playing at a I’m playing at a much more basic level, so there’s less to figure out and it’s more obvious.
31:26
Okay. Okay. I I I’ll give you that. Um,
31:32
let me jump into another technical question here because I’ve I’ve read a little bit of a dissertation on how you
31:37
execute this and you and I have had a previous conversation about this and I know that this implements MIDI very very
31:45
heavily and um I’m a longtime MIDI uh been
31:51
involved in it since the early days um and I know its
31:58
strong suits I also know its limitations, especially MIDI 1.1 or whatever we want
32:05
to call it, the old MIDI as opposed to 2.0, which has definitely uh done a lot
32:11
for the genre, so to speak. But, um, in terms of using MIDI,
32:19
does that I mean, yes, you can use MIDI in a very human way. All right. But
32:24
aren’t you then sort of um translating it to and from again? I
32:32
mean, how does that how does that impact the actual humanity of the performance?
32:38
That’s a good question. And and that also came down to uh sort of desperation
32:44
on my part for another gig that I was I was hired for. [snorts] Um, and it was a track with John Ferraro
32:52
and Sterling and Chad Wackerman, uh, and Jim Cox and Blues Sakarano and Steve
32:59
Morse. But Steve wasn’t originally on the track.
33:04
It was just blues soloing in this section. And they wanted to extend the song and have,
33:13
you know, Steve play a solo, which ordinarily you would just loop it.
33:18
But with players like those and you know, especially drummers like John and
33:23
and Chad, they’re going to be completely playing off not only Steve, but each other
33:30
because both drummers were playing at the same time. Yikes. Okay.
33:37
And I realized that the only way I could do this was to construct a performance from scratch like I did on Live at the
33:44
Z7. But to do that, I needed to be able to
33:50
convert everything to MIDI, be able to change the MIDI, and then when you rerender the MIDI, it’s exactly
33:59
what the audio would have been.
34:05
Okay. Can you go into ju just for for others who are hearing about this for the first time, can you go into a sort
34:12
of cursory explanation of how how rendering it into MIDI impacts the
34:19
process? Sure. And I can show it to you too. Cool. Even better.
34:24
Uh there are two things that happen first. One of them is to figure out exactly
34:30
what they played. And that’s al also the first problem because there are some really good AI
34:37
score but they’re not perfect. This has to be perfect otherwise it’s
34:44
not the same performance. Well scoring in general it has its limitations you know. I mean, it’s
34:51
something that was created to, you know, as a as a friend of mine likes to say, you know, the fly on paper is not
34:57
going to necessarily be exactly what they played, which is another issue, which is how do
35:05
we capture every part of the performance into MIDI? Right. Right. How do you capture the
35:11
nuance that doesn’t get translated into a into something written or data
35:16
generated? Well, the data gets split. So the the note data
35:25
is converted to MIDI and then another process
35:31
learns about the instrument and it doesn’t just take like a snapshot
35:37
of the tambber uh like like MIDI veil does. It actually figures out the
35:43
physics of the instrument and how that instrument makes sound and how that instrument works.
35:50
And this would be a specific instrument, not just guitar, but the guitar of Steve Morse, for example,
35:56
on that track specifically. Uhhuh. Because if you’re going to change what that person played,
36:02
you have to know how to reynthesize the notes that would correspond. Because
36:08
when you play a note quieter on a snare or a guitar, it does not sound the same,
36:14
right? Nor does the the note that comes after it.
36:20
So those two things happen in parallel and Prism generates a custom virtual
36:26
instrument just for that track. So if it it converts it to MIDI and then
36:33
it creates the virtual instrument and if you don’t edit the MIDI, you get exactly
36:39
the original track back. How do you get the audio exactly the
36:46
same? Well,
36:52
it’s a lot of stuff. Um, [laughter] it’s hard. It’s a very hard problem.
36:59
Yeah. But but you you understand why I’m asking that, right? I mean, I can sample a guitar. I can pull up a million guitar
37:06
sample libraries and load them into MIDI. How do I get exactly what Steve
37:12
would have played or someone else for that matter would have played? Well, that that’s where that’s why it
37:19
took me a long time to uh to develop this. And unfortunately, I did this in
37:25
2018. So being able to use, you know, things
37:31
like transformers and LLMs was nent
37:37
challenging. Yeah. Uh yeah. So uh but there’s an even bigger problem
37:46
and that is if someone hits a crash symbol say Marco Minimman great drummer
37:52
hits a crash symbol and then 143 milliseconds later he hits a
37:59
different crash symbol. Now these symbols are only recorded in the overheads.
38:07
Well, that means my system has to reach in and pull out those symbols and the isolation has to be perfect
38:17
and all the ambience has to be removed perfectly.
38:24
So you have these two symbols. Let’s say the second symbol you want to move it
38:30
this way. Well, that means you would have to hear
38:36
the decay of the first symbol, which wasn’t originally generated,
38:43
right? It was the sound was so masked by the sound of the second symbol that it
38:49
didn’t get recorded.
38:55
And and I would argue uh that it’s not going to be captured just by those overheads because you’re going to get
39:00
ambience in the rest of the kit, too. Right. That’s right. That’s very important.
39:07
[clears throat] So, this system has to be able to figure those things out and it has to do it in
39:14
real time and it has to do it with phase accuracy
39:19
and it has to do with that’s a lot of math. It’s a lot of math. uh with what I call cognitive accuracy,
39:26
which was something that you and I talked about, which was the feeling that it was part of the performance, which
39:33
doesn’t happen when you fly in a a sample. Yeah.
39:38
And I can show you an example of this. This is a program called Spectral
39:44
layers. I know it will. Ah, well,
39:49
when I was working on that the second PhD,
39:55
I realized I needed to be able to do spectral editing. And I found this amazing spectral editor called Spectral
40:01
Layers that was languishing at Magics.
40:08
Yes. And I was very fortunate. And I got in touch with Robin Leel, who’s the the
40:14
developer of it, and I said, “Why don’t we collaborate on a new version, and instead of an application,
40:23
we’ll do it as an Arra plugin that replaces the waveform editor, and I’ll
40:30
sell it to Steinberg, making,” [laughter] which is what happened. Yes. Um, and so
40:38
we we got to work and Robin is the most extraordinary audio software engineer I
40:45
have ever encountered. And in a year we reimagined and re and
40:52
rebuilt with a new interface uh and a new engine uh what became spectral layer
41:00
6 for Steinberg. And what I was using it for
41:07
stuff like this.
41:28
So here we have an overhead symbol track which is the green
41:34
And this blue border here is one crash symbol. And this blue bookmark here is a second
41:42
crash symbol. And between them is 143 milliseconds.
41:47
So here’s a overhead symbol [music]
41:57
and so fast it’s hard to hear both of them. It’s just
42:03
Yes, but I do hear. Well, what would that symbol have sounded like if it was completely
42:09
isolated and he hadn’t hit the second symbol
42:23
H. [clears throat] Now, keep in mind this was in 2018. This
42:29
would be a lot easier to to do today. Sure. U now the question is how do we know
42:36
that that’s what the symbol would have been?
42:41
One way is to play both at the same time.
42:49
And not only should there not be any phasing, there just should be that symbol getting
42:55
louder.
43:04
[music] So there that the symbol is now so loud
43:09
that it’s masking the other symbol, right?
43:15
[music] But otherwise it sounds exactly the same.
43:26
Okay, another thing that you can do is if you look at the waveform
43:36
up close of the sec of the overhead track that
43:42
was recorded. You should be able to see the telltale
43:48
vibrations of the first symbol vibrating the second symbol. And it
43:54
should match exactly the waveform of the isolated symbol.
44:03
And you do. H
44:23
Okay. So, h yes, I see this. Now, let me let me
44:32
zoom out to the lay person’s question again and say,
44:37
why do I want to do this? Why do I want to pick this performance apart that deeply? What am I
44:44
accomplishing? Well, when it when I did it, it allowed
44:49
you to take all of your audio tracks and only have MIDI tracks,
44:55
which Ah, okay. Okay. So that you can then isolate each
45:01
component of the audio track and turn it into a more accurate representation via
45:08
MIDI. Yeah. And all the problems of processing audio go away. And you can
45:15
use all the tools for MIDI but for audio
45:20
it makes everything much much easier. I will show you. Okay.
45:29
Okay. Um now let’s bring this back around.
45:34
Oh go ahead. finish your explanation first and then I’ll then I’ll fire off my next question. I
45:40
I was going to give a a demonstration of using it in a in a DAW, but uh yes, you
45:46
want me to do that? Yes, please. Okay. So, here is a
45:53
uh a session uh the drum session and let’s say
45:59
I want [clears throat] to do high hat strikes.
46:06
P Studio One is not giving us audio now. What to do?
46:12
Oh, excuse me. It’s not called Studio One anymore, is it? No. No, it is not.
46:20
You got that? Yep.
46:26
So, what we can do, I’m going to take this. I’m going to put it on the track
46:32
now. The high hat track, original high hat track was encoded into Prism’s database.
46:40
And then as were all the drums, we can load up the high hat and we choose the
46:46
articulation we want, which is strikes.
46:54
Then we generate the tracks.
47:02
And when we do that, it also created the virtual instrument.
47:12
[clears throat] So in this MIDI representation, [clears throat]
47:19
it’s a staircase because each note corresponds to a
47:24
specific note that happened at a specific time.
47:31
So notes are used not for uh pitch
47:37
but for choosing which particular hit in the original recording.
47:44
So each hit becomes its own its own unique sample. Yeah. Except it’s not really a a sample
47:52
[clears throat] because you can’t do things like move them around if they were samples. Move
47:58
them closer together. pull them apart, make them louder, make them softerware uh softer.
48:06
So if we’ll
48:18
just bring this in.
48:32
So if we listen to the beginning here,
48:39
listen to just the MIDI. [music] Listen to just the audio track.
48:48
Let me do that again.
48:53
Uh listen to just the audio track.
49:02
And then just the MIDI track.
49:14
It’s exactly the same. And if you look at the audio, if you render the audio, the white is the audio from the MIDI,
49:22
you [clears throat] can see just how similar
49:29
it is in terms of recreating what the original was.
49:37
So wait a minute now MIDI is zeros and ones. What is the
49:43
audio source that the MIDI is then creating or triggering or
49:49
it is a virtual instrument which was assembled by listening to the
49:57
original it listened to the original high hat track and for but it’s not samples you said
50:02
not samples. So then what is it that’s creating this
50:08
audio? Where what is the source of the audio? is it it’s it’s not a recording of
50:16
of the original audio and it’s not a I mean it’s not a sample. So what is it
50:22
that I’m hearing when I’m listening to the MIDI because I’m not listening to MIDI data. I’m listening to a MIDI performance.
50:29
So where’s the audio coming from? Well, it’s being reynthesized by the virtual instrument.
50:36
So the virtual instrument is able to read the MIDI and based on what note it
50:42
is, it knows exactly what that note should sound like and can reynthesize it
50:49
in real time. You can also add new notes to the track
50:54
and it will figure out what that note would have sounded like if it was performed right there based on the note
51:01
that came before and the note that came after. and reynthesize that as well.
51:08
So, so this virtual instrument is uh is
51:14
it a standard virtual instrument that I can buy somewhere uh or is this a
51:20
virtual instrument created uh uh is it a proprietary instrument that’s created in
51:26
this process? It’s a proprietary instrument created in this process. Uh, Prism right now is
51:32
just something that I use with my clients to solve uh, problems. It’s not
51:38
a commercial thing, but it works with drums, piano, uh, guitars, and bass,
51:45
and it allows you to just get rid of all your audio tracks, put it all into MIDI.
51:50
But there’s another reason that I’ve been pushing on this lately, and that is because of generative AI.
51:59
And my concern that a lot of people I think share obviously is is that what
52:06
place do real musicians have when anyone can generate a performance
52:13
in the style of soand so or yeah that was going to be one of my questions.
52:19
And so the last few years, what I’ve done with Prism is turned it into
52:25
something to allow musicians to be on par with generative AI to give
52:33
them new abilities so they can be competitive. So as musicians, we stay relevant
52:41
because we can do things that AI can’t do.
52:47
And it means that DAWs, which
52:53
are probably going to go away if something like this doesn’t really if a change like this doesn’t happen in the
52:58
industry, there’s no reason for DAWs and effects and virtual instruments and all
53:04
of these things. But if we can give musicians
53:10
superpowers that AI can’t do, then we have a
53:17
fighting chance. What is the one thing you’re a musician?
53:24
What is the one thing that AI can’t do?
53:31
it [sighs and gasps] at I I I feel like this is a trick
53:38
question. Not a trick question. [laughter]
53:43
I I mean I know that in my in my view what AI can’t do is recreate my exact
53:52
performance. And that’s why you’re scaring me here. Well, what you’re hearing required your
53:59
performance in the first place. Mhm. [clears throat] Right. Because then it’s essentially it
54:06
it’s using artificial intelligence to create a model of my performance.
54:13
Mhm. And I I think you’re right that the the one thing AI can’t do,
54:19
and this is for mathematical reasons, not because the technology doesn’t exist. It’s mathematically impossible
54:26
for the AI to be you. Mhm. It could generate stuff that is
54:33
plausible to be you. I’m talking about, you know, just a completely new song,
54:38
but it can’t reproduce what you would have played in that moment,
54:44
right? It can’t do any of that.
54:51
And so if we take the ability of musicians to be expressive,
54:56
personally expressive, and we allow musicians to do that easier
55:04
and to give them new ways to do that, then our expression becomes more
55:09
powerful. Prism does not generate audio
55:15
and it won’t it’s not designed to generate expressions.
55:20
But once your performance becomes MIDI, you can do you can change your mind
55:28
about things that you’ve done. You can make very small adjustments very easily.
55:34
There’s always a distance between what we imagine and what we actually
55:40
play. Oh, absolutely. And Prism allows you to close that gap
55:45
really easy. It’s literally just a question of
55:51
being able to move the MIDI notes around and and change the the other
55:57
specifications of that.
56:03
I’m not sure how I feel about this, Bill. I mean, I I mean, I I I see
56:10
I see that this is a lofty goal. My question to you would be, what do you
56:16
say to somebody who says, for example, the imperfections are part of it. I
56:24
mean, you’re allowing me as a musician, you’re empowering me to essentially
56:29
perfect a performance. You’re empowering me to go back and secondguess what I
56:35
might have done, what I would like to do, um what I would what I wish I had the chops to do, for example.
56:42
So um you know in that respect
56:48
okay hypothetically what would then stop me now I I you know
56:54
I understand that you you probably have a lot of protections in place here in terms of what you [clears throat] will
57:00
allow and what you you know what ethically is possible but let’s toss aside the ethos for a moment and say
57:08
given this technology what would stop me from loading in a Steve Morse profile
57:16
and playing something really really sloppy and and and
57:22
more like I would play it and then say convert this to a Steve Morse style for
57:27
example. Well, the difference is that in prism that model is licensed from Steve Morse.
57:35
Oh, I understand that. I understand that. However, what I’m saying is that
57:41
as with all things technologybased and particularly with AI, the fear among
57:48
artists and creative in particular is that this technology will then start to
57:55
quote unquote escape. You know, we have we have virtual models of, you know, Morgan Freeman wishing you happy
58:02
birthday and all of that kind of stuff. Um, what what would you say to somebody that
58:08
says, you know, this eventually could jump the shark and, you know, all of a
58:14
sudden, even though it’s proprietary for you right now, what would stop somebody
58:19
from eventually getting to a point where we can kind of fake it, so to speak?
58:25
Well, I I I I think it did jump the shark. the the the very first time this
58:30
was uh ever done to take a performance and make it in the style of someone was
58:37
those seven seconds of Steve Morris’s solo and then you know 15 years later it
58:43
could make Morgan Freeman. But in that example that you gave of you can make
58:48
your solo sound like Steve Morris, it’s not going to fit what you wrote.
58:56
The only person who can make that fit is you. If Steve Morse played that solo, it’s
59:04
not your song anymore. And from one perspective,
59:10
the process of making music is something that people enjoy.
59:17
Exactly my point. But is are we taking this away? Well, I think we’re we’re we’re preserving it
59:22
because [clears throat] one of the things that,
59:27
you know, when when Sooner really became powerful, I had an awful week where I said, “Why
59:33
should I even write songs? No one’s going to know what I did.”
59:38
I’m having that argument right now with an artist that I’m working with who is brilliant. She’s exceptionally talented,
59:44
but she’s leaning on Sunno. And one [clears throat] of the things that I pointed out to her in our last
59:49
conversation was in much the same way that if you allow everybody to live with
59:56
the demo, the rough mix for long enough that becomes the mix. Well, if I allow
1:00:02
you to live with these generated track, does that then become the track that you
1:00:08
want that your brain wants? That’s happening to me on an album I’m doing right now as well. I’m running
1:00:13
into the same thing. Yeah. And Yeah. And and you know that’s so that’s where I start to get into the you know kind of ethical you know
1:00:22
creative argument here of is this doing harm or is it doing good or is it a
1:00:27
little of both? Well, you know, when Dune Technologies are
1:00:34
created, there are a lot of there’s a lot of concern casualties.
1:00:40
Like I’m think I’m guessing that when MIDI was in created, there were a lot of
1:00:45
people saying, well, that’s the end of musicians. Oh, absolutely. And we can say that about synthesizers, about drum machines,
1:00:52
you know, we we were both there for both of those, you know, and and I totally agree with you and I I have been a a
1:00:58
firm proponent for many years stating unequivocably that, you know, if you
1:01:04
take that technology and put it in the hands of a hack, you’re going to get crap. But if you put it in the hands of
1:01:10
a true creative, they’re going to figure out how to then make that into another instrument.
1:01:16
You know, I mean, the most creative artists are gonna get hold of anything at all and just figure out how they can
1:01:22
break it, how they can do something with it that wasn’t intended, how they can make it into something. and and I don’t
1:01:28
argue with you that the that the potential for this is astronomical,
1:01:36
but I but having also, you know, now living with all of these AI arguments of
1:01:42
what’s happening when people allow Sununo to create our songs and stuff like that. I just wonder about that
1:01:51
particular, you know, that that ramification of it of the whole idea of um
1:01:57
of basically, you know, music being a form of human expression. Are we are we are we
1:02:05
with that? I I think that we are. You know, I just be, you know, just because you’re
1:02:10
paranoid doesn’t mean that you’re correct. Just because [laughter] every other time we’ve thought, “Oh, no.
1:02:17
this is the end. [clears throat] That doesn’t mean at some point there isn’t the end. Sure.
1:02:22
I that’s why I’m so passionate about prism is because
1:02:29
you can only create really good music if it is personally expressive of you in
1:02:37
the moment. You can have the most advanced AI create
1:02:44
something really cool sounding. It’s not going to resonate with people
1:02:49
emotionally and it won’t be fulfilling for you as a musician because you will
1:02:56
not have had the creative process of these emotions flowing out of you into
1:03:02
these vibrating strings and people can tell
1:03:09
authenticity. Absolutely. Absolutely. I agree with you and I’m you
1:03:16
know I’ve always been a firm believer that you know there’s a lot of mystery involved in the creative process and uh
1:03:22
you know it’s been it’s been posited many times that
1:03:28
we don’t create these things they are created through us you know that there’s this creative unconscious that basically
1:03:34
the ideas just filter through us that’s why a lot of times you know a song will just kind of pop out and then you know
1:03:40
you’ll look at it a week later and go wow did I write that you know so I mean there’s There’s definitely, you know,
1:03:46
there’s a lot of that and I and I totally agree with you on all of that. Um, but I’m I’m playing devil’s advocate
1:03:53
here and saying, you know, there’s a lot of people who are going to look at this and say, “Whoa, danger, Will Robinson,
1:03:59
because it looks like there’s an aspect of this that really can sort of
1:04:05
shortcircuit the creative process.” And you know there’s there’s a you know
1:04:11
there’s a danger of uh shall we say unhealthy use of this you know.
1:04:17
Well you may be right. I hope not. Uh especially and I’m not believe me I I love it. I
1:04:24
really do. And I’m I’m not you know I’m not picking it apart to be pedantic or anything like that. Um I’m just looking
1:04:30
at it and I’m saying wow this is it’s so powerful that it’s kind of scary.
1:04:37
Well, to to I’ve been running a background process in my brain about a question that you asked a little while
1:04:43
ago and you said this gives people the ability to create something that’s
1:04:51
perfect, but I you know that ship already sailed with Sunno.
1:04:57
Absolutely. And and it it sailed long before that sailed with quantization, you know, and
1:05:03
Yeah. So if you want to use prism to make
1:05:09
yourself sound perfect, you can.
1:05:14
If we are interested in preserving human expression, well then if you use it to try and make
1:05:20
your sound perfect, you will not be succeeding in your music and it will be lacking something. So, you can make that
1:05:27
decision, but beyond fixing some things that just didn’t
1:05:33
weren’t how you wanted, the further it gets from what you actually played, the less emotional
1:05:40
resonance it’s going to have. Okay? And I I say that with with scientific uh
1:05:48
science to back that up. Uh there is one thing that I I really liked
1:05:56
about it that may not be obvious and when I you know I’ve been really fortunate to work with the all these
1:06:01
virtuosos and what I found absolutely
1:06:08
surprising was that none of them are happy with their performances.
1:06:14
That doesn’t surprise me at all. If anything, they’re going to be more hyperritical than than we, you know,
1:06:20
meat and potatoes musicians would. Even Mike, who’s like the most confident
1:06:26
musician I’ve ever met, and all the time that we’ve worked together, I’ve never heard him say, “Ah, I did some that was
1:06:33
really good what I just did, you know?” Yeah. Sure. And
1:06:40
Steve explained to me that when he plays,
1:06:45
instead of focusing on just what he wants to express, he also has to focus
1:06:51
on not making a mistake. Hm.
1:06:58
Because if he does then well that will just you know he has to do another take
1:07:04
or it’s going to get protoolled which means it’s you know notes are going to get copied and you’re going to lose the
1:07:10
whole performance. A performance has to be start to finish. If you you start overdubbing parts and it it
1:07:18
doesn’t feel the same. I agree 100%. Yeah.
1:07:23
[clears throat] The prism allows you to just think about the performance and not worry about mistakes
1:07:32
because afterwards you can fix it with so much precision that you’re not
1:07:38
autotuning it, you’re not quantizing it. You were making it sound oh that thing
1:07:45
where I made a mistake here, here’s how it would would want wanted that to have sound. So you keep the flow of the
1:07:51
performance and I think you end up with better performances because I know it’s
1:07:56
the same thing for me when I’m when I’m tracking. I’m just like don’t screw up, don’t screw up, don’t screw up and not
1:08:02
really thinking about but the problem is that before prism
1:08:08
I sound like a like a paper towel commercial. Uh but before prison, [laughter]
1:08:14
the tools to fix these issues were so blunt that they
1:08:20
ruined the performance. Absolutely. Absolutely. And that, you
1:08:26
know, that’s been part of, uh, you know, one of the flags that I wave a lot of times, which is that, you know, now that
1:08:32
we have so few limitations, you know, we don’t have we don’t have a a limited
1:08:39
number of tracks and we don’t have, you know, we have have every plugin imaginable that musicians no longer
1:08:46
bring their agame. you know, it’s not a matter of like it it’s always a matter of, you know, oh, I can comp it or I can
1:08:52
do that again, you know, and there’s never this this continuity of a single
1:08:58
performance and that to me has always been I mean I understand that it’s just a different way of doing things and you
1:09:05
know I can adapt to both but you know I kind of every once in a while I sound
1:09:10
like oh those were the days when I start talking to my students about yeah you know when you only had two tracks to do
1:09:16
a vocal you know, you were much more diligent about really what you were
1:09:22
doing. And I think this also kind of feeds into that, doesn’t it? I think so. I I learned that lesson hard
1:09:30
when we we mixed an album with Michael Brower and and I showed up with 256
1:09:35
tracks. Oh my god. And Mike said, “You can have 48 tracks.”
1:09:45
And I said, ‘Well, no, because like there’s all these, you know, decisions that he’s like, that’s
1:09:51
exactly right. You need to make decisions. Yes, you need to commit. And that’s one
1:09:58
of the things that limitations I mean there’s a great book uh it doesn’t have anything to do with the music industry
1:10:04
per se but there’s a great book that I read many years ago called the power of limits and it it talks about exactly that how
1:10:12
the more the more we can postpone committing and making decisions the less
1:10:19
we’re going to be able to do later on because we’re going to have this what I refer to as the paralysis of analysis.
1:10:26
Yeah. you know, absolutely where where we’re literally, you know, and and look,
1:10:31
mad respect to, you know, I remember seeing this video of Phineas editing uh
1:10:36
one of his sister’s vocals, you know, down to the syllable. And, you know, mad respect to people who can do that, but
1:10:44
that’s a big difference from, you know, Ed Stasium telling me that, you know,
1:10:49
Glattis Knight did the entire performance of Midnight Train to Georgia in one take. Yeah.
1:10:55
you know, and and that, you know, they’re I mean, obviously there’s a lot of middle ground in between those two,
1:11:02
but the idea that man, you know, a performance is a performance. And you
1:11:08
know, you could argue that back in the era before we could record things, a performance was ethereal, you know, and
1:11:15
you you had to be there, you know, and we’ve we’ve gone steadily further and
1:11:21
further away from that with our ability to do more and more.
1:11:28
And so, you know, what you what you are creating here is both brilliant and
1:11:34
scary. And you know, you’re right. I mean, it really depends on whose hands you put it
1:11:41
in. I absolutely agree and I’m fascinated. I, you know, I’m like I’m
1:11:46
the first one to say, man, you know, my left brain wants to dig into this. Seriously, you know, and and play and
1:11:52
frolic around in in in this little in this world. Um, at the same time, you
1:11:59
know, it’s definitely there there are challenges there that need to be addressed, I think. Are there not?
1:12:06
you know, I hadn’t thought of some of the things that you brought up and I would love to, you know, work with you a
1:12:13
bit on some more feedback from you because I’m hearing the things that you’re
1:12:19
saying and I’m like, “Oh, you know, I didn’t really think of that. I didn’t think of
1:12:26
this thing where it could paralyze people with uh with choice.
1:12:35
The the example I like to use is the Slate Trigger. It’s a wonderful piece of software
1:12:41
that’s got five. It’s got five microphones that you can adjust and change. And I’m
1:12:49
like two two hours into a track and I’m still thinking, man, you [clears throat] know, maybe I could get that fourth
1:12:57
microphone [laughter]
1:13:04
Yeah, it’s it’s um Oh, and I’ve been there myself, man. You
1:13:11
know, um I mean, I had a conversation with one of my students yesterday where we were talking about the whole idea
1:13:16
that um you know, how much time have I spent trying different snares? You know,
1:13:23
how much time have I spent going through 150 reverbs, you know? And yes, there
1:13:29
are some instances where nuance is important. Absolutely. You know, does
1:13:35
this particular instrument is this particular instrument, is this particular preset, whatever, is it what
1:13:41
the song needs? But ultimately, if we get too deep down that rabbit hole,
1:13:46
we’re going to get so caught up in the process that we’re going to lose sight of why we came here. And this is a you
1:13:53
know this is a big um this is actually something that I have done um panels on
1:13:58
at AES at NAM uh we’re about to do one very soon uh with the simpy folks about
1:14:04
the same thing you know how how deeply do we get caught
1:14:11
the way I phrased it to somebody the other day is remembering why we came in the room in the first place
1:14:18
you know we came here to make music man we came here to create art and it seems
1:14:23
that a lot of times we get wrapped around the axle of the technology
1:14:29
and so I’m you know I I I see where this can be an absolutely brilliant tool and I also see where it could create even
1:14:36
more of that cycle of you know analysis paralysis. So it really you know a lot of it really
1:14:43
depends on how you employ it. I guess you’re right.
1:14:50
I think there are also some other uh yep to uh be able to
1:15:00
advocate on the other side. There are other benefits that are not
1:15:05
obvious and the most striking one to me was what happened when I went to mix drums
1:15:13
because the first thing that usually happens in modern production is we replace the drums.
1:15:20
[laughter] Yep. And I did some cognitive uh analysis
1:15:26
and found out that your brain reacts differently with when samples are dropped in because
1:15:35
the expression the brain is very very sensitive to the
1:15:40
flow of expression like like you were saying and when you pull in samples
1:15:48
you lose that and and In fact, Michael uh Mike and I were talking about this uh recently that as
1:15:55
soon as you you drop that in, you’ve lost the performance.
1:16:01
And I think it’s subtle, too. I think, you know, maybe consciously we don’t realize why it’s disturbing us.
1:16:07
Yeah. Well, it but the the thing that’s that I find challenging
1:16:14
in that is if you don’t do that, your drums don’t sound good. because we’ve never figured out how to record drums.
1:16:22
Yep. And then I realized we we did um a new Steve Morris uh band album and Van Roma
1:16:31
I asked him about the drums that he set up and he said, “Yeah, I spent about two hours, you know, picking out exactly the
1:16:38
drums to fit the song.” And I said, “Well, how do you feel about whenever you do that, an engineer goes in and
1:16:45
replaces them?” So, not only are you losing the expression,
1:16:51
you picked out those drums and you tuned them. That would be like if you played a Stratacastaster and I was like, “Okay,
1:16:57
well, we’ve replaced that with like a Gibson tuned to drop D.” Now, no other musician would
1:17:04
allow that, but drummers sort of have to.
1:17:10
Well, and then, you know, you could take that even a little bit further and, you know, dig down into the nuance and say,
1:17:16
“Okay, maybe I’m not replacing the drums, but maybe I’m putting some heavy compression on the snare, or maybe I’m
1:17:22
EQing the toms.” And immediately now I have affected the sound of those drums
1:17:27
that you’ve labored so hard to get so perfect. You know what I realized
1:17:32
was like 90% of when we do that is to make up for problems in the process of
1:17:39
recording drums. Yes. So when I brought up Prismact and I could have a slider for each No, for
1:17:46
each instrument [clears throat] and the tambber the full Yeah. I didn’t realize that ride symbols have low end
1:17:55
or that they could make a sound other than just a ping. But if you want to mix a record, it’s
1:18:02
going to be a ping. Otherwise, you got, you know, and I just brought all the faders up to zero.
1:18:10
And this the mix was done. I did a little bit of EQ, but I for the
1:18:16
first time I didn’t have to layer anything on top of that snare because when I had this just the snare by itself
1:18:23
and I had the um articulations isolated. So prisms can say just turn up the ghost
1:18:29
notes. Uhhuh. When I first did this with John Ferraro,
1:18:35
he said this will play change the way drums are played. And I said why? He said, “Because
1:18:41
there’s so much we don’t even bother doing because we know it won’t end up on the record.”
1:18:47
He said, “All the subtlety in our playing is stripped away as part of
1:18:53
modern drum drum production to make the drums sound big and full and
1:18:59
and have a lot of energy because of all the problems in recording the drums.”
1:19:06
Absolutely. And then, you know, that brings to mind another another variable. And I’ll play devil’s advocate for you
1:19:12
here again, too. What do the drums sound like to the drummer versus what they sound like in front of the drummer?
1:19:19
Yeah. You know, and you can say that about any instrument really, you know. Um, but
1:19:25
okay, maybe an amplified instrument, you know, I’m going to be standing in front of the amp and so are you. But if I’m playing an acoustic guitar, you know,
1:19:32
what’s it sound like from up here over it as opposed to out there in front of it?
1:19:37
You know, and I mean these are all variables that once again, you know, when you talk about the whole idea of as
1:19:43
soon as you take on the idea of capturing the performance,
1:19:49
you know, putting a microphone or a bunch of microphones, even worse, in a room, you’ve created a whole new set of
1:19:57
variables. Yeah. So, I mean, and and you know, and we I mean, we could chase this tail around
1:20:03
forever here because there’s so much to it, but um you know, again, and I
1:20:08
realize what you’re talking about here with Prism is yet another tool. It’s a
1:20:14
damn cool one, and it does things that no one has really done before, and I’m
1:20:21
I’m fascinated by so much of it. Um, and [clears throat] and yet,
1:20:26
yeah, man, it brings to mind so many questions. And, uh, yeah, I I’ll tell you honestly, you know, both on and off
1:20:33
record here, man, I’m I’m I would be eager to uh to implement some kind of uh
1:20:39
some kind of working together on this, especially with my university students. You know this is like man because this
1:20:46
is you know one of the things that I have been really fascinated about for
1:20:51
several years now is the idea that those of us who came up in you know the
1:20:57
dinosaur era of uh you know I mean I started out with a fourtrack recorder
1:21:02
you know and so for me I have watched not only the technology change but the
1:21:10
change that the technology has implement the changes that that these technological evolutions
1:21:17
have made in the creative process itself. And that to me is much bigger
1:21:23
than the than the changes in the technology itself. I mean, you know, it’s like it’s great that I can now
1:21:28
carry an entire studio in my MacBook. That’s awesome. You know, I love that.
1:21:34
And I love the idea that I can have multiple instances of a compressor that
1:21:40
I used to have to, you know, really pray that I could afford one of, you know,
1:21:45
and and stuff like that. And you know, I’ve got every synth imaginable that I pined after when I was, you know,
1:21:52
15year-old, you know, um, agonizing over whether or not I could afford a Mini Mog
1:21:58
and instead, you know, having to buy a a Sequential Circuits Pro one cuz it was 500 bucks cheaper, whatever, you know,
1:22:04
all of those kind of limitations. And now I don’t have those limitations at all. And I have students who have never
1:22:13
had those limitations. And so for them, the creative process is
1:22:18
180 degrees out from what I grew up with. How does that impact things? You
1:22:23
know, how does that impact the way they write a song? How does that impact the way they orchestrate a song? And this is
1:22:30
something that it’s literally an ongoing day-to-day conversation that I have with
1:22:36
younger people because, you know, the creative process, the need to create is
1:22:42
still the same. It’s still a human a basic human need and yet we now have a completely
1:22:47
different pallet of tools and you know this again this is another quantum leap what you’re doing here
1:22:54
and uh you know that’s going to definitely impact things just the same way that bringing in a drum machine did
1:23:02
or bringing in a synthesizer or unlimited tracks or plugins or any of that stuff you know and so yeah how are
1:23:10
we going to use this you And and what I love is the idea that we can take this
1:23:15
now and we can put it in the hands of very masterful people,
1:23:21
you know. I mean, as I said, you know, I mean, you can you can take you can take
1:23:27
this kind of technology and put it in the hands of people who who kind of suck
1:23:32
and they’ll kind of suck a little less, you know, which, you know, and that’s fine, you know. Look, m music
1:23:39
should be access to suck a little less. Me too. Absolutely. You know, I mean, I
1:23:45
I look, I would love the idea of, you know, there’s a bass part that I’ve been
1:23:50
wrestling with now on one of my own songs for for weeks and weeks. And it’s like, you know, just sitting down and
1:23:55
playing it right through the whole thing, you know, as opposed to trying to punch in and stuff like that. And I
1:24:01
would love it if something could just come along and say, “Oh, no, dude. Here’s what you really wanted to play.” You know, that’s great, you know. Um,
1:24:08
and because in my head I can play it, you know. I would love if there was a technology where I can say, “Okay,
1:24:14
here’s the baseline I want.” And just hum it and have it then generate my part
1:24:20
the way I would have played it if I wasn’t fumble fingered. You know, all that stuff sounds really cool and that’s
1:24:25
the kind of stuff that I’ve, you know, I’ve wished for my entire musical life. You know, it’s like, you know, you you
1:24:32
you have these dreams where you’re playing, you know, you’re playing keyboards like freaking Tchaikovski, you
1:24:38
know, then you wake up and realize that you can do triads on a good day if you’re lucky, you know, [laughter]
1:24:45
but but you know, at the same time, I think there’s there’s some real validity there of if I can take this and I can
1:24:53
put this in the hands of a real artist, you know, and as I said, you know, one of the things I love about technology
1:25:00
ology is giving it to people who don’t have a preconceived notion who haven’t RTFMED who can just say hey what happens
1:25:07
if I you know and and believe me I’m I’m a guy who used to write manuals okay so
1:25:13
I know you know it’s like no I know that somebody’s going to come along and figure out well if I push all four
1:25:19
buttons what happens then you know [snorts] I mean all that kind of stuff man that’s great and I I’m excited about
1:25:27
what this can do in the right hands Um, but you know, again, I just I I just
1:25:34
know that I’m spending a lot of time having discussions about the ethical and
1:25:40
less than ethical aspects of AI. And so, the fact that you’re digging into this,
1:25:45
you’re going to run into those questions obviously, and I’m sure you already have. Well, yeah. I But I you really put an
1:25:53
idea in my head that I really hadn’t considered. Now I’m thinking about what kinds of things I can put in prism that
1:26:01
push people toward making decisions so they do we do make better music. So
1:26:08
this thing doesn’t just give me the tools but the workflow has a creative
1:26:13
pull toward what’s going to be most effective because we can do that. If we break it
1:26:20
down to discrete events, we could do anything with the workflow, which means, you know, I’ I’d love to
1:26:28
explore what that would look like. You know, what would the system do to uh
1:26:34
pull people toward that? I’m in. Absolutely. This sounds this is
1:26:40
fascinating to me, man. Um, well, I don’t I don’t want us to get get too
1:26:46
long-winded on this whole thing here because I know that we could make this into a 4-hour episode really easily. Um,
1:26:52
and you know, we can certainly continue this, but um, you know, let let’s find a
1:26:57
way to sort of wrap this conversation and and um, you know, I will I will
1:27:03
promise my audience here that this is going to be an ongoing conversation because this is I mean, we ain’t done
1:27:09
here by any means, man. This is uh this is obviously something that uh there’s
1:27:14
so much to explore here and I’m I’m fascinated by it. I really am. And uh
1:27:20
I’m I’m excited for the possibilities here and I’m excited to actually try different things with this now that I
1:27:27
have a basic understanding of how it’s going to work. And u at the same time I
1:27:33
see that, you know, there’s going to be some interesting guardrails to put into place here. Yeah, I’m realizing that now.
1:27:40
You’re right. [laughter] Um, well, let let’s let’s wrap this. I
1:27:46
know you also wanted to talk about some some uh some new surprises that you had in mind for some of this stuff. Do you
1:27:53
want to get into any of that? Well, uh I was going to demonstrate um some of
1:27:59
the new uh new features. Um one of them is called the time machine.
1:28:06
So, I really got into this idea that anything that you do artificially to the
1:28:12
sound takes away from the performance. You know, if Steve wanted all the notes
1:28:18
to be the same volume, he would have played it that way, right?
1:28:23
But there you may want some type of dynamic
1:28:31
control or something with the room. Another thing that that that Michael
1:28:36
Brower taught me was that and Jay Graden was that if you put the microphone in the right place
1:28:42
in the right room and you set it up correctly, you don’t need to do any engineering.
1:28:49
Yeah, Al Schmidt showed me that. So I said, “Well, how do we get to that?”
1:28:55
Because it’s also for someone who’s starting out, the idea of these really complex effects,
1:29:02
I think takes you out of the creative process because they don’t feel musical, they feel artificial.
1:29:08
Mhm. I said, well, like heavy compression, for example. Yeah.
1:29:14
What if you could go back in time after a recording was made and change any
1:29:22
aspect of it? you could change what the instrument
1:29:29
was, maybe, oh man, I wish we’d used this particular snare. So, instead of
1:29:35
layering it, well, you know, what if we’d used that snare? Or what if we tuned it like this? Or what if we’d used
1:29:41
these microphones? Or what if we recorded it in a space that had these characteristics?
1:29:49
Well, and a lot of that has been done to a certain extent.
1:29:54
you know, with drum replacement, with uh microphone modeling, convolution,
1:29:59
reverbs, all of that kind of stuff. It has, but there it hasn’t been done in
1:30:05
such a way where sort of in one control panel, you can just choose what you want
1:30:12
to turn back. And there’s a very unique type of drum replacement that Prism does
1:30:18
where it doesn’t replace the drum. It only replaces the resonant chamber.
1:30:25
The everything that happens with the stick hitting the head is preserved.
1:30:30
Otherwise, there’s no reason to you’re just sample replacing that. There’s no point to
1:30:35
right. But this allows you to keep every aspect of the performance. So, it really is
1:30:42
exactly what it would have sounded like if it just had been a different drum.
1:30:49
Okay. Okay. So, we’re preserving all the nuance of the player. We’re just changing the other circumstances.
1:30:57
And you you have this virtual room thing where you can just change and it’s oriented. It’s funny. If you look in the
1:31:04
Prism plugins, there are very few knobs
1:31:09
because of this thing about I don’t want to give people choices. I want to give people three knobs that let them do what
1:31:15
they want. And so the entire time machine interface is like five buttons and knobs for all
1:31:24
the parameters. There’s you can’t there’s nothing in the
1:31:29
Prism plug-in where you can finally adjust something because I that’s not
1:31:35
like you said that’s not really useful. Uh so
1:31:41
yeah I I’d love to show you that uh at at some point the time. Mhm. Well, actually,
1:31:46
yeah, I need to show it to you real quick. We won’t I do audio, but I’ll I’ll just uh
1:31:52
Where are we here?
1:32:03
Here’s the performer section. Uhhuh. And
1:32:09
this is sort of an example of just some of the the the crazy things that you can do
1:32:14
uh when something is missing. And actually, let me open a different
1:32:19
one. That’s better.
1:32:26
So, this allows you to time your notes based on when we perceive notes to begin
1:32:34
the beat, not when they actually begin. Because the timing in every MIDI sequencer is wrong because the bar is
1:32:42
not when we feel the beat, it’s when the the note begins. But in prism, whenever you see
1:32:50
the beginning of a note on the piano roll, the note actually starts before that. And the edge, the left edge of the
1:32:58
note is when we feel that the beat begins.
1:33:09
And I I mentioned that, you know, we can get away from a lot of the problems that
1:33:15
we have had with with audio. And so if you wanted something like a perfect compressor,
1:33:22
because it’s MIDI, you can assign anything to anything. So if I want to assign, I don’t know, velocity to
1:33:27
amplitude. Uhhuh. I can draw a curve like this.
1:33:33
And this is a perfect compressor. It has no it doesn’t change the sound. There’s
1:33:39
no attack. There’s no release. You don’t need those things. This simply and perfectly without affecting the sound
1:33:46
will give you this velocity curve. Interesting. And it’s that easy to do.
1:33:53
And there it is. Activate. Wow. Now another thing is because Prism
1:34:00
knows every note that’s in the performance instead of a traditional mixer where you have to assign
1:34:07
instruments to faders, you can it will assign notes
1:34:15
to faders. So there are three uh symbol hits on this track. So there are these
1:34:21
three note faders because Prism models the entire decay of
1:34:29
the note. If I make this note louder, even if this note came right afterwards,
1:34:36
it’ll still sound natural because this note will continue to decay and make this note vibrate.
1:34:45
So it’s going to impact the notes following it, right? You won’t just hear
1:34:50
and so like if I make this note soft and this note is loud, you don’t just suddenly hear something come in and then
1:34:57
this thing getting cut off. Each note is re is rendered individually
1:35:04
and then put together. H and so I found
1:35:10
Okay. So, so you are really you are really getting closer and closer to creating a virtual
1:35:16
reality, so to speak. Yeah, it uh
1:35:25
I’ll just go to the high. So, here’s all the high hats on this track. And you can just draw that and that’s, you know,
1:35:32
that’s it. That that’ll be the the volume. And you can if you want to have attack and release and decay, you can do
1:35:39
that for each note, which would be for, you know, if you had
1:35:45
one problem note, you know, you could do that or you could adjust the velocity
1:35:50
response for any note again individually
1:35:55
because it’s MIDI, right? Now we get to the time machine.
1:36:04
And again, you’ll notice that there are no parameters for like adjusting.
1:36:09
There’s just making the thing happen. So you can tune the drums after the
1:36:15
fact, which doesn’t seem that big a deal until you realize if you try that on a real drum kit, even if you had an AI
1:36:23
algorithm that removed all the bleed, all the drums are also in the overhead
1:36:28
symbol. Yes, that’s the big problem that that you run into with any kind of live drum
1:36:36
kit is that anything you do to one mic, you’ve got to do to them all, right? But in Prism, there are no overhead
1:36:42
mics. You get rid of them. [laughter] There’s no need because every instrument
1:36:47
is on its own channel. If you want the original reverb at the end of it,
1:36:53
you can do that. Here we can add new mics.
1:36:59
We can move the mic and it’ll do that with phase locking.
1:37:04
So, ah, you know, okay, you know what I mean by that. Uh, here you can adjust the
1:37:10
tension on the head. And here you can choose different symbols.
1:37:15
Mhm. And then for the ambience, if you want to restore the native track ambience, you can. If you want to put that
1:37:24
ambience on a bus reverb and process everything through that ambience,
1:37:30
say, “Okay, I want the ambience from the overheads.” But it’s going to extract that ambience profile. And there it is
1:37:38
at the volume it was recorded. But, you know, if you want more or less of that ambience,
1:37:43
so I’ve essentially put my entire kit into a different room at that point. Sure, you can do that, too. Like right
1:37:49
here. or you can do it three-dimensionally. Now, to do 3D audio, you have to remove
1:37:57
100% of the ambience from the original recording, but we’ve done that. So, you can just literally have X, Y,
1:38:07
and Z, and that’s what the sound does in space. Uh-huh. Sure.
1:38:12
Or you can have an anooic chamber. I mean, yeah. Now you’re getting into basically like a convolution kind of a
1:38:19
thing. Yeah. essentially. That’s right. Uhhuh. But of course, you have to extract that
1:38:24
uh perfectly from the uh original and you know the right these things keep going. Um
1:38:31
one of the things that that I really liked was with the drum modeler that if
1:38:37
you have a tom that sustains for a period of time, well that’s how long the MIDI note is. You can keep dragging that
1:38:44
slider and the tom will just sustain very naturally
1:38:50
because when you’re mixing drums, the duration of something makes it
1:38:57
louder. So you can do that instead of actually turning the amplitude up.
1:39:03
Right. Right. So you can make the kick drum go louder. Um you can change the shape of the note.
1:39:11
Um you can also here you can change things like how much ring is in the snare.
1:39:17
Uh how much um uh see what it has here different
1:39:24
characteristics of it. How much ring there is um how much resonance there is in the body. Uh but again it’s
1:39:32
oh I I know a lot of engineers are going to love you for this. But notice it’s also there’s just two knobs and I’ve
1:39:39
just picked out two things and each one has a knob. That’s it. Can’t do it. You can’t change anymore. And that’s a
1:39:46
deliberate design decision. Otherwise, you you’d never get anything done.
1:39:52
Right. Right. Huh. Yeah.
1:39:58
Like I said, there there are so many variables here and and some of them are great. As I said, engineers are going to
1:40:04
love you for a lot of this, for being able to uh get rid of a lot of the issues that we spend hours and hours on
1:40:12
when mixing a track, you know, because let’s face it, we all spend way too much
1:40:17
time on the drums. Yeah. You know, [clears throat] and this
1:40:22
definitely this definitely addresses a lot of that. uh it gives you a lot of power
1:40:31
and and yet it doesn’t really I mean what I what I’m fascinated by here is that this is the first approach I’ve
1:40:37
seen where you’re not really changing you’re not replacing the drums.
1:40:42
Why do that? Drummers spent two hours and how many years learning how to do
1:40:47
ghost notes and and high hat foot chicks and then they just get stripped out.
1:40:52
That’s why people say three musicians and a drummer because they all the the
1:40:57
performance is in pop music is taken out because you need to to mix the drums.
1:41:04
Right. And we end up making it all sound like a drum machine. Exactly. And and that has always been, you know,
1:41:11
one of my biggest beefs is because I I appreciate a really really good drummer.
1:41:17
And in a sense, they’re kind of rare. You know, they’re more rare than I wish
1:41:23
they were. And that’s why we end up spending a lot of time quote unquote
1:41:28
fixing drums. And and I I emphasize quote unquote
1:41:33
because a lot of times what we’re doing is we’re taking the soul out of them. Yeah. on on on this particular track. Um
1:41:42
there’s Marco Minimman was was uh playing and he was the track had a certain
1:41:47
[clears throat] feel and then I noticed that if I really
1:41:53
looked at the high hat track there were these foot chicks. They were so quiet because of the sound
1:42:00
of everything else. You could couldn’t even really tell sometimes that they
1:42:05
were there. But I put it through Prism and it just gave me the foot checks
1:42:11
and it turned out it’s interesting that you mentioned Marco because I I um I was literally at
1:42:17
the side of the stage watching him play at Montro many many years ago.
1:42:22
And yeah, he’s he’s a really really nuanced drummer. I mean it it’s such a
1:42:30
such a pleasure and yet you know I can see where that would be really a challenge to not only capture in a mix
1:42:38
but bring those subtleties out so that they stand out in the mix. That’s right. And here you can literally
1:42:45
just push up a slider. [clears throat] And that’s I guess part of why I think
1:42:51
that this will I want Prism to let musicians become the
1:42:57
best version of themselves, not someone else, not something else.
1:43:02
And part of that is people being able to actually hear your performance better.
1:43:08
It turned out that Marco was playing a 16 beat on the foot chicks and on each
1:43:16
uh on the down beat he would flip his foot around so that it made a different
1:43:21
sound. So he was playing a whole drum beat on and it was never heard. And when
1:43:27
I turned that up so I could hear what he was hearing, it was a whole different drum performance and the whole song
1:43:33
changed. Wow. Yeah. Well, you know, you just you just
1:43:40
said something that I think really it sums up a lot of this, which is uh
1:43:47
you know, wanting musicians to become better versions of themselves rather than somebody else. And that
1:43:54
right there, I mean, if you can create that,
1:43:59
then that is a huge huge feat right there. And I think really just a
1:44:05
wonderful thing. I mean, that’s what we all want, I think. You know,
1:44:11
I don’t want to I don’t want to play like someone else. I want I might want
1:44:17
their influence on my playing, you know. Um I mean, I I remember um I remember
1:44:24
watching Marcus Miller during soundcheck at Montro one year
1:44:30
and I was literally I was standing there at the edge of the stage and he was he was playing with his band. He was
1:44:35
playing I Want You Back. Okay, which is, you know, that’s not an easy baseline by any means, right? You know,
1:44:42
and at the same time, he was actually singing it like Michael, man. You know, just all the nuance. Oh, baby, give me
1:44:48
what? You know, the whole thing, right? And he finished, he looked down at me and said, “Hey, how you doing?” And I
1:44:54
said, “Man, I’m about to go home and burn all my bases, right?” [laughter] And you know,
1:45:00
and you know, he was totally cenial about, man, he was great. you know, he was like, “No, man. You just got to practice more.” And I said, “Dude, I
1:45:07
could practice like, you know, for the rest of my life, and I’m never going to sound like that, you know, but the point being, I don’t want to play like Marcus.
1:45:14
I would love to be able to have some of the dexterity that he has. I would love to be able to to play some of the nuance
1:45:21
that he has, but really, I would want that as bettering my own playing.”
1:45:28
And you know, that to me is really, I think, what what most self-respecting
1:45:35
musicians would want. You know, don’t don’t don’t let don’t help me play like
1:45:40
Hrix. Help me get Hendrickx’s influence into my playing. Yeah. Which is different. It’s a
1:45:46
different thing. It’s a completely different thing. And it’s, you know, it’s the difference
1:45:51
between, you know, being your own musician versus sounding like someone
1:45:58
else. You know, I remember years ago there was a guy, um, what was his name? Randy Hansen. Everybody said, “Oh, he
1:46:05
sounds just like Hendrickx.” And I’m thinking, “Yeah, but Hendrickx already did that.” [laughter]
1:46:10
Probably Randy. He’s not even trying to do that. And he’s like, exactly. And I I felt [laughter] I kind
1:46:16
of Right. And I kind of felt for him. was like, “No, dude. You know, you probably want to be respected for the musician
1:46:22
you are, you know, and and that I think is really what sets us apart is that we
1:46:30
want to be unique. Each of us wants to have something unique about ourselves.”
1:46:35
And that’s difficult, especially in an age where everything is evocative of
1:46:42
something else because we’re all playing off of our influences and we’re all filtering our influences and everything.
1:46:48
So, so anything we can do to help refine the artists that we already are,
1:46:56
I think is a great it’s a great endeavor. Well, you know, you could, [clears throat] if I’m lucky, you can
1:47:02
help me do that better than I’m doing it because that is exactly what I’m trying
1:47:07
to do. It It’s exciting to me, man. Definitely.
1:47:12
Um Wow. Well, um I think we I think we’re
1:47:17
definitely going to need a part two of this bill. Okay. Um you know, and and definitely I I want
1:47:23
to hear more about this. I want to um I would love to be uh to have some involvement in this because this is
1:47:30
fascinating to me. And I think especially from not only from an educational perspective but from the
1:47:36
perspective of someone who um like yourself is old school but also works
1:47:41
with people who are not you know and people who have never known those limitations.
1:47:47
Yeah. I think this is uh there’s there’s a lot that could be there’s a lot that could
1:47:53
be achieved here. Well, thank you. That means a lot to me coming coming from you.
1:48:00
Well, thank you. It’s it I’m I’m fascinated by this. Uh I really want to
1:48:05
I I want to dig into this some more, man. Definitely. Um well, uh any any
1:48:11
concluding thoughts here? Um what do you what do you see as the future of all of this, man? What do you where do you see
1:48:18
us going with this technology? I see
1:48:23
a divergence between music that’s created um by AI completely
1:48:32
and this idea of musicians that are focused
1:48:37
on performance and that’s what their fans are like and that’s what they’re interested in and more of a move toward
1:48:44
live performance and the idea of a recording instead of being you can
1:48:50
create anything in the studio, but people are going to want to hear things that are more stripped down, more
1:48:56
authentic, and more like uh a live performance. And that we’ll have sort of
1:49:01
a a repeat of something like the grunge movement, which I think was a a reaction to how computerized and controlled and
1:49:09
artificial music was in the 80s. Absolutely. And you know, same thing
1:49:15
like punk. Yeah. You know, I mean, there there there always is a reaction and a response to
1:49:21
stuff that’s too clean. You know, it’s it’s funny. I was uh I
1:49:27
was chatting with a friend of mine and we were talking about uh there had been um a radio station in LA that had done
1:49:36
sort of a mashup of a number of the current hits duour.
1:49:41
Oh, yeah. It was an FM station. It was doing the classic rock stuff. not classic rock, but the uh the modern rock kind of
1:49:47
stuff. And uh they literally were going from one to the next, same mix, same
1:49:55
almost the same key, same BPM. It it you know, there was no real subtlety in it.
1:50:04
It was like formula. And you know, you when you start doing that, you’re going to have a reaction. You’re going to have
1:50:09
push back. You’re going to have people who, as you say, they want to go out and they want to hear a real band,
1:50:16
you know? They want the genuine article. Man, I’ve I’ve noticed that uh
1:50:22
there are more virtuoso players now than ever. I can’t believe how good these
1:50:30
young players are and how incredible their technique is. I listen
1:50:35
to a lot of um modern progressive metal and there’s just so many incredible
1:50:42
players and it just makes me feel so
1:50:47
optimistic about, you know, the current, you know, the young generations because
1:50:52
they’re obviously already drawn toward this idea of
1:50:58
the these of performance and and actually being able to play, you know,
1:51:04
like Nice. I got a lot of those people in my classes, man. You know, just amazing players, you know, people who I I mean,
1:51:11
I’m excited to go back into, you know, school starts again in in a month and I’m excited because I’m thinking like,
1:51:17
oh, I got this track I want so to play on, you know, because, you know, really, you have so many musicians who are truly
1:51:25
just amazing musicians and they’re just looking for a platform. Yeah.
1:51:30
I I think that, you know, like Sabrina Carpenter when she’s on stage, like
1:51:37
they’re actually playing music, right? You know, and that’s just Yeah.
1:51:43
It’s not trash. Yeah. Yep. And so I’m I’m really optimistic about
1:51:50
what’s coming, but I also think that there is a lot of danger um to mitigate.
1:51:58
Uh so yeah. Amen, brother. Absolutely. Yeah. Yeah.
1:52:05
Um well, we’re we’re coming up on two hours here, so I’m going to I’m I’m going to try and find us a good uh a
1:52:12
good way to conclude this thing. But uh as I said uh this is far from over, man. This is this is too fascinating for me.
1:52:19
And uh and I know that a lot of my audience is also really really interested in stuff like this because
1:52:25
man, this is this is where we’re headed, you know, and it definitely is. Uh it’s
1:52:31
different. It’s different from all the run-of-the-mill stuff that we’ve done before. And u man, I I applaud your
1:52:38
efforts. Um, I think you’re absolutely patently insane in a good way. And
1:52:45
[laughter] I’ve heard that just I mean honestly, man, the idea that
1:52:51
you are pursuing this is just like holy dude. Really? Um, you know,
1:52:57
[laughter] it it truly is amazing to me and it’s fascinating and I’m I’m excited to see
1:53:03
where it’s going to go and uh I like I said, anything I can do to be a part of
1:53:09
it, man, I’m I’m excited because this is, you know, as with many of us, you
1:53:15
know, as a musician, I am I am firmly divided between left and right brain.
1:53:20
Yeah. You know, and uh this kind of appeals to both.
1:53:26
And it’s one of the very few things that does because you know for the most part as musicians
1:53:31
when we get to a certain point with our art it becomes rightrained.
1:53:38
You know we stop thinking about the technical aspects of it. We stop thinking about the the the
1:53:44
motor motor functions and stuff like that. And so this is this is a bridge.
1:53:50
It really is. You know it’s a bridge between those two. And that’s uh that’s an interesting place to be.
1:53:58
You know, I’ve realized that, you know, we were talking before about the value of having to make decisions.
1:54:05
And every part of prism was because of something I had to do or
1:54:11
I didn’t have a decision and if I hadn’t needed to make that
1:54:16
decision, it never would have happened. So I think Prism itself is an example of
1:54:23
what you’re saying about minimizing choices and instead focusing on well
1:54:28
what did we come into the room to do? Yeah. Yeah. I love that, man. That’s
1:54:36
that’s why we’re here. That’s Yeah. That’s what that’s what started us down this path, you know.
1:54:44
Yeah. That’s fantastic, man. Well, Bill Evans, thank you very very much for
1:54:49
being my guest and for for this time and uh I’m really looking forward to
1:54:54
continuing the conversation and uh we will talk very soon.
1:55:00
Thank you so much for having me.
1:55:06
Hey, I’m Daniel Keller. Don’t forget to hit the subscribe button and join us each week for Insights and Sound.