Sam Fischmann - The TRUTH About Mastering in Modern Music Production (AKA: It's OK To Master Your Own Music) - Progressions: Success in the Music Industry

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
Let's get rid of the technical hurdles so you can spend a lot more time
on the creative hurdles. That's what we want to do as musicians, as producers, as
mixers. That's Sam Fishman, the software developer musician behind the audio plugin
company, Music hack. Since Sam makes a mastering plugin, I thought it would be fun
to get his perspective on the role of mastering in modern music production,
which we obviously got into, but we also got into so much more than that.

(00:20):
Everything from the problem with our obsession with loudness targets.
Once something becomes a metric, it gets abused, and people pay more attention to the
metric than the reason the metric was invented.
Actually be thinking about loudness. You will hear different things
in the loudness. If you like it, that's what matters. Yeah. You're not
gonna damage it by submitting a track that's too loud. We also get to the

(00:42):
role of AI and machine learning in music, including some of the potential
copyright violations. Tech companies could be making. The ethical barrier there
is what aren't they telling us about the training data. And that's something that
I believe the artists should have a right to when it's being used
for that kind of thing. We just don't know. And a reminder that regardless of
the tech, regardless of the tools, you have to trust your taste and make

(01:04):
music that you like. What really matters when you're trying
to finalize a track and make it really pop and everything is how it sounds
to you. Just like cooks need to taste their food. If that's not the number
one thing you're paying attention to,
no one can help you. So if you're a mastering engineer, mixing engineer, artist,
doesn't matter. I think you're gonna enjoy this one. So stick around. For my interview

(01:24):
with Sam Fishman.
We were gonna talk about the role of mastering in modern
music production, and I thought that I kind of
knew what it was. I mean, I still think I know what it is, but
I did this episode recently called is Mastering Dead? And it
was a commentary on the value of working with a mastering engineer that

(01:48):
complements your sonic and musical tastes instead of using
AI. One button. And
the comments on that video, not that comments bother
me, but they were, like, overwhelmingly negative towards
paying for mastering. And it shocked me. It made me feel like I live in
this bubble of just the LA recording industry, where we all

(02:09):
just. We followed the path. And so you make a mastering tool
that is great, and it's laid out for musicians and producers to
use easily and get good results. What do you think the role of
mastering is in 2024. It's a really good question.
There's, like, so much to talk about, so I gotta rack my brain here and
figure out where to start. I. There's. There's two things I normally think of that,

(02:31):
like, immediately when people ask me about mastering. Well, do
we need to talk about what mastering is for some people? Because a lot of
times people think about mastering, it's like a dark art. You know, they're like, well,
what is it? I mean, it. I feel like the.
I feel like the veil is being taken off every day when, like, a new
tool comes out. But, I mean, I think mastering something different, everybody.
Yeah, yeah, that's true. Like, and I have a lot of, like,

(02:54):
friends and relatives that aren't in music, and they ask me what I do, and
I'm like, all right, am I making plugins? And the first plugin is a mastering
plugin. And depending on what they do, I use a different analogy. So if there's
someone who likes to cook, I tell them, like, it's like getting your
seasoning right, like getting the salt right. All of the creative decisions about making the
dish have, like, been made. You know, what are the ingredients? What's it gonna taste
like? Is it gonna be cold or hot? Like, what are the textures? All the

(03:15):
stuff. Right? But if it doesn't have the salt right, it just doesn't pop. And
if you have too much, it just tastes weird and artificial. And if you don't
have enough, it tastes blande. But you're not trying to change the whole thing. You're
just trying to get it. Just, like, in this space where all of a sudden
everything comes together. And then for people who like cars, I'm like, well,
you can clean a car all you want, but until you wax that thing, it's

(03:35):
not going to shine and be ready to be put up on a pedestal, you
know? I don't know what it's like to over wax a car, though. So
you can definitely do too much when mastering. I don't really know if you can
over clean a car unless you, like, you know, start rubbing the paint off it
or something like that. But you can do that when you're mastering because you can
always do too much. That's true. That's true. Yeah. So that's kind of the
overall level of it. But when I break it into two parts, there's two things.

(03:57):
There's, like, the. There's, like, the creative aspect
of mastering where it's all about your taste. You mentioned that, like, you need to
find someone that aligns with your vision for your music vision,
your. What's the word for vision? But with
your ears, I don't. Whatever
your imagination of, like, what you want your vision. Right. And then there's the

(04:19):
part that's, like, the technical. The technical part of
mastering. And I think that's.
That's where there's, like, a big divide, I think, in sort of the
criticism of paying somebody a lot. If you had comments. I didn't read the comments
in the video because I'm afraid of YouTube comment sections, quite frankly.
So, like, if what you're looking for in mastering

(04:41):
is a second set of ears to just to do that
polish, like, I need somebody who I trust that, like,
is aligned with me and will catch those little things or, like,
really lean it in a direction and does something that I can't do
with my inner ear, with my skill set,
in terms of my taste or my. That

(05:04):
part of it, your vision, that is something. I
don't think that can be replaced, honestly, because you
can't. I mean, another person and someone who you vibe right with,
that is what it is. But the technical side of it, I think, is that's.
That's the part where we're trying to rip down all the barriers and be like,
no, you don't have to have a million dollars in gear. No, there aren't 50

(05:25):
million technical targets you have to hit. No, you don't have to individually phase
align drum hits so that you can get the maximum headroom out of them. No,
no, no. All of those things are like, yo, like,
you're just being sold stuff at that point. And I make a product, right? So,
like, this is my pitch, in a way, when I say these things, right, which
makes me feel a little bit. But. But I really want people to understand that,

(05:47):
like, what really matters when you're trying to finalize a track and
make it really pop and everything is how it sounds to you. Just like, cooks
need to taste their food, right? And if that's not
the number one thing you're paying attention to,
no one can help you. I agree. I agree. Is that
the phase alignment? Interesting. Like, commentary on

(06:09):
that, you know? Cause I'm an engineer. I'm, like, in there. Yeah, I get a
bunch of kick drums from somebody. The first thing I do is look at the
phase and it'll be tighter. It'll be like,
there'll be more thump and whatever, and then I'll flip back and forth between the
rough mix and there's, like, this, especially in, like, some of the pop
styles, it's kind of cooler the way that it
was. And you're like, I fixed it, but I broke it, you know? And

(06:31):
these rules are meant to be broken. They're like guidelines, you know? Yeah.
And there's also, like, this comes up in all sorts of
ways in life, but, like, once there's. I forgot what the words goals
are corrupting, I think is. Or something. Like, once something becomes a metric, it
gets abused. And people pay more attention to the metric than the reason
the metric was invented. Right. In work, this happens because,

(06:54):
you know, you just cheat to get the numbers looking good, so your boss gives
you a raise or whatever. But, like, in music, it might be, oh, I want
to hit a luft's value. Oh, I need to maximize the amount of volume I
get out of a peak. Oh, I need to whatever. And if you keep a
narrow focus on this one thing, you're gonna do it to the detriment of other
things, which is why you always have to keep an ear for the whole thing
anyway. How does this relate to what mastery means in

(07:17):
2024? And I think what that means is if you
have the money to and, you know somebody that you trust, and you really
need that second set of ears because. Because
it is that extra opinion, that humanity, that gives you an extra little
touch on the music cause another person give it a second pass,
I think that's always a thing. I agree. But

(07:40):
if you're worried that you don't have the technical skill, but you
do trust your taste, that's becoming less and less of a thing, and I
don't think it should be a thing anymore. I think that if you know what
you're listening for, there are tools available to you that are not too complicated. I
make one of them that will get you what you want,
and it's really just about listening and

(08:02):
knowing what it is you need to change using reference tracks.
I talked to Bob Yasinski a while ago, and he said, why aren't people bringing
up listening to reference tracks when they're doing a mix? And I think I
dodged the question because I wasn't expecting it, but it's 100%.
If you're making music that you want to sound like something, listen to the thing
you want it to sound like. And if you know

(08:23):
that's what you do as a producer and a mixer, or whatever part you're doing,
you're like, I gotta lean it a little bit brighter or a little bit tighter,
or whatever it is you need. I agree entirely. But I want to go back.
Okay, sorry, sorry. Like, there was something that you said that really
caught me, so we'll just go off on a tangent. The
thing you mentioned, I can't remember what you said, but it was essentially that, like,

(08:45):
once a standard is invented, people kind of abuse that standard without thinking
about where that standard came from. Right. And that is such an
interesting statement because it really, like,
take the -14 Spotify normalization standard.
Right? Yeah, yeah, yeah. It was. Every day there's a
conversation with somebody about how loud something should be,

(09:08):
but nobody talks about why that is there.
I'm literally, this is what my talk on the talk that I'm going to give
it at the audio developer conference in November is called in praise of
Loudness. And I'm specifically gunning for this. And
what's also really interesting, if you read Reddit, I try not
to, like, post much on Reddit, but I read it a lot. There's, there's a,

(09:30):
I think it's, I don't remember if it's mixing, mastering, or if it's, like production
or something. And they have a sticky at the top where one of the moderators
or someone who's, like, in good with them wrote a whole, a whole essay
and pinned it to the top about this very topic. And it's like,
read this before you post anything. We're tired of hearing about
it. And it's, and it's, it's cool. It's well written, I

(09:51):
think. I haven't thought about all the parts. I agree and disagree with it, but,
like, the general gist of it is like, I'm glad it exists. And almost was
like, do I really need to give my talk now? But, but,
yeah, that's a good one. Do you want to talk about that much, or are
you just bringing it up? I kind of do. If there's pieces of that talk
you want to practice, do you want to. Hear

(10:12):
where that number comes from? I do. Yeah. Okay.
So the original issue is simply the,
like, you know, some things are loud and some things are quiet. Like the
classic dilemma of I'm watching a tv show and then an ad comes on and
it's three times as loud. You're like, oh, my, I gotta turn this down. Right.
But the same thing happens with music. Right. Especially

(10:33):
now that we have playlists that are mixes of tracks from different
albums. And it was a common complaint of
users of various music listening platforms. Not just
that, but broadcast television, movies, all sorts of stuff. Right, right.
And so there were various committees that did a bunch of research to try to
decide, well, how do we. How do we solve this problem? It sounds simple on

(10:55):
its surface. You're just like, all right, let's. Let's just, you know, let's look at
blur. No. Okay, everything's fine. Right? But they had to do a bunch of
studies because there's a bunch of little problems that crop up when you try to
do this. In fact, can you guess what the original number they came up with
when they're like, all right, we're going to standardize on a lufs level. This is
after they invented Lufsden. Can you guess what the original lufts level? They thought that

(11:15):
ever all media should be normalized on. And Ebu still
says eventually we should get to it. I mean, tv is like 23 or
24, right? Negative 23. Yup. Yeah. But they made an
exception for music. And the reason they made an exception for music, which was negative
16, we're still not at negative 14 yet. Okay. The reason they made an exception
for music is they're like, okay, there's a problem.

(11:37):
The converters on these cheap players suck.
So if we're shipping music at negative 23 lufts and people have, like,
crappy headphones and a crappy player, or even they bought a good player, but they
used a crappy DAC, you are going to have to amplify it
in order to get it to a listening level that you like. And if it's
crappy and you're amplifying it a lot and you have a bad signal to noise

(11:58):
ratio, you're just pushing a lot of noise into that line, and it's not a
really great situation. So they're like, well, we got it. We got to make music
louder so that people that are listening on, like, portable players or whatever, I don't
have to crank the cheap amplifier on their
portable phone or on whatever they're listening to. Right? So negative 23 was not
acceptable, but then they had even more problems because they're like, oh, you

(12:19):
know what? We realize that different tracks on an album are different
levels, so we need to come up with a way of
normalizing the volume that keeps the relative differences between
the tracks. And they're like, okay, well, if we take the
loudest track on the album and we normalize that to
-14 so there's a little bit of give on either side,

(12:40):
then we take the value we got, let's say, okay, it
got pushed down by three decibels,
then we just do everything relatively the same. So the things all stay relatively the
same distance from each other. So we have track
normalization, album normalization, media broadcast
normalization, all this stuff. Right. But the real thing that's

(13:02):
scary is the misinformation that people got from this negative 14
lufts. That's how we got to negative 14 lufts. And by the way, that's not
even standardized kind of. They could change at any point. It's not really a
standard Amazon. That's why if you listen to watch movies
on different streaming things for video, sometimes you have to turn the volume way up
on your app, and sometimes you don't. They're not the same. And who knows what

(13:23):
they'll be next year. So trying to hit these targets, it's, like,
not gonna help you. Also, let's say you submit a track that's too loud.
Why I say too loud? I just mean louder than negative 14 laps, which isn't
too loud. There's nothing wrong with that. Look at all the pop music. It's all
mastered. It's negative eight. Yeah. So let's say you do
that. Oh, no, they turn it down.

(13:43):
That's what we call a linear process. It's the same as turning down your volume
knob. There's nothing. It doesn't distort your audio. It's
pristine. It's just quieter. That's fine. Yeah, right. Yeah. But
imagine. Think about what happens when you turn audio up. What if you submit a
track that's, like, way too quiet and they decide to change the
standards? You'd have to put a limiter in there. That's exactly right.

(14:05):
The peaks start to go over and you have to put a limiter. So the
danger is that you're submitting tracks that are too quiet. They
can't push them up because they don't want to put a limiter on them because
it'll change the quality. Some do, like, there's a mode in Spotify for,
like, louder normalization, and they will put a limiter on the track if they
push the volume up, but it doesn't by default. So it's just like. But who
knows? Who knows what they'll do next year, right? What if people complain? The whole

(14:28):
thing started because people were complaining. So the real danger is submitting
tracks that are. That are too quiet. Here's another one. Nobody talks about this.
There's also this whole true peak thing, right? Yeah. And then they also talk
about needing peak headroom because,
oh, if we convert to a lossy format, then you're
going to cause more phase problems. That's going to create your peaks. Going to push

(14:49):
higher. Oh, it'll hit a DAC, the headroom thing. Well, it'll pop.
First of all, Dax already have headroom. If they didn't, they'd be terrible
all the time. So that's supposed to be their responsibility.
But here's, here's the real kicker. Let's say, let's say that you're worried about two
problems. One, they're going to lower the volume of my track because it's too loud.
Oh, no, we already talked about why that's not really a problem. Right? Two, if

(15:11):
they're lowering the volume of your track at eight decibels by like six decibels
to get it down to negative 14, then why are you worried about peaks?
Do you have six decibels of headroom? What's one db gonna
do if they're gonna lower you? It's just like it does. They don't, they don't
even match each other. If you're following all the guidelines, you're doing a bunch of

(15:31):
stuff that changes the characteristics of your track to please
somebody who's gonna change their mind at any time and whose advice, if you follow
all of it doesn't even add up. They make sense in vitro,
but not together. And, and people will say, like, oh, well,
you can't control the final device. Right? So
the thing that converts, as it turns out, MP3 s, they don't know about

(15:54):
peaks. They store information in a way that does not clip.
The clipping happens when the MP3 is decoded. Right. So
if the MP3 is clipping or the AAC is clipping,
it means that the decoder is crappy. They
should normalize it after decoding, and they should have a decoder that decodes into 32
bit. None of that should be your responsibility as a producer or mastering engineer.

(16:18):
Are you doing this so that, like, TLC and Vizio can save $2 a unit
on tvs? You know what I mean? Like,
why, that shouldn't even be the purview of,
you know. Yeah. Like, it's crazy. It's crazy to me. What you should be
listening to is if I like my mix, I like how it
sounds and I dial the volume up and I like how that sounds because when

(16:40):
you push things louder, even if you take a lot of control of the transients,
like, you will hear different things in the
loudness if you like it. That's what matters. Yeah.
You're not going to damage it by submitting a track that's too loud. No, you're
not going to damage it. If you have peaks that aren't insane, as long as
they're within, whatever, give yourself a db of headroom. Give yourself half a

(17:01):
db. You can always convert it to an AAC and listen, and if you don't
hear anything, it's fine. I really think
these conversations have been damaging to people because they're getting to focus on all sorts
of stuff that has nothing to do with how your music sounds. And that's
bad. Yeah, no, I agree. That was a big rant. No, no,
I love that rant. That's perfect. I

(17:22):
think that lays it out in a way that most people don't
lay it out. Especially talk about the true peak thing is interesting, right? You're like,
okay, I need to have 1 db headroom, but then they're going to turn it
down seven decibels. So I don't need that one db headroom when I'm mastering
because they're going to turn it down. But, yeah,
that one will mess with you. It reminds me of early

(17:44):
Mfit mastered for itunes days where it was like,
I mean, let's talk about something that was like a bit of a
scam. Like everybody in the beginning that I knew was mastering for cd and then
turning it down until it passed the little checker. And then you're like, mmm,
mfit. Here you go. Yeah.
So, yeah, but is that for your music or is that for the

(18:06):
distributor, but the distributors?
Well, there's a whole other conversation you're talking about, like, who's in charge? You
want to talk about atmos and spatial and whatever, but we won't go
there. No, I think that's great. I'm glad that you shared that
the way that you did because people are
really obsessed with the number and I've always been.

(18:29):
I use this argument when I'm talking about atmos a lot and it's like, you
have to make a thing that will live for eternity,
right? So if that thing sounds killer at minus six lufts
somehow, congratulations. If it sounds
good at minus eleven, that's where it sounds the best. That's okay,
too. Like, you have to make a product that will translate to

(18:51):
all of the future formats, not worry about what the number is this year
for. For some companies. So true. You know, spec
sheet. So, uh. Yeah, that's good. Love that. That brings up another thing.
It's not just all the stuff in the future, it's also the current. Like,
it does matter. The current listening environments, not just the.
Not just the technology you're listening through but the spaces you're listening in. And this,

(19:14):
I think, is also another conversation that leads into the
overemphasis on dynamic range as
a positive in and of itself, which isn't
true. As in, more dynamic range is more. Positive, as
in more dynamic range is always better. Exactly. It's not. That's
not the game. It is true that dynamic ranges can be great, but,

(19:37):
like, one of the reasons you can never listen to classical music in the car
is because the dynamic range is too much for you to listen to it in
that environment. And is that actually the feeling you want when you listen to
all music? No. No, definitely
not. No. We listen to music. We
listen to music. Headphones in loud environments. We listen to music

(19:59):
in the background. We listen to music in the car. We listen to music at
the gym. We listen to. There are so many places, and we want it to
sound great everywhere. And that's part of what mastering also does.
And the trick that mastering engineers have developed, not the.
But one of the tricks that they developed over, like, so many years,
is how do we make it louder? How do we

(20:19):
technically reduce the dynamic range without it sounding like
we're reducing the dynamic range? Yeah, because then you can hear it
better and it doesn't feel worse. And that's the
game. Yeah, but you can't measure that, because
all you can measure is what the dynamic range is and what the loudness
is. You can't really say

(20:40):
it sounds. It sounds really dynamic, but it isn't. Is
not really something we have a measurement for.
And that's why these guys are revered. Guys and gals.
Sorry. The best mastering engineers I've ever worked
with. It comes back just
so loud, but yet not.

(21:02):
You can't tell. You're like, well, how did he. How did he or she
get this to this volume and make it sound so
punchy, not distorted? Like, this is. The people that can do
that are worth every penny because they have figured something
out. Yeah. Okay, here's the. Here's the thing. Here's the thing. So
what's that? Is that. Is that a personal

(21:24):
vibe thing or is that a technical thing? That's a technical thing. That
is something that tools can solve. That is something that can
be made democratic by people having access to the tools that can
do it, because people that, like, understand the process and have built the tools
can do it and make it easier. So when you're trying to increase the volume
of music, that isn't a vibe's decision as much as, like, the

(21:47):
feel, the width, the breadth, all these other things that are less tangible.
Yeah. So that is the part where it's like,
you can do that with the tools on the market now. I mean, that's one
of the things our tool does best. So that's why I'm so confident about
it. But that's separate from the part of, like,
do you need someone who's gonna give you a vibe check? Yeah, yeah,

(22:08):
yeah, totally. Before we move on to a
couple other things I have in my notes, I want to, going back to those
standards, right, the -23 where is that
standard? And the cd loudness
wars, where do those things line up? Cause that, there's no way cds were
at -16 they were always going to be louder than that. Right? So,

(22:30):
yeah, cds were always loud. The 23, I forgot, I don't have it in front
of me right now, but I was going back into these. Eb,
what's the name? There's. The AE's is the American engineering society.
They do some studies and they have some stuff. And then there's,
I forgot the name of the european, I think it's EbU.

(22:50):
European Broadcast Union. Maybe we'll make that up. Somebody will comment on that.
Yeah, Ebu also does these studies. And so the negative 23 came from a
bunch of research, I believe, that was done by the EBU. And they were basically
saying, like, what level can we get stuff on that
we can standardize for literally all media? It has enough

(23:11):
dynamic range built into that level where
you have all sorts of stuff you can do with cinema, but you could also
provide music at it. I'd have to look at the paper again to remind myself
exactly how they centered on that number. But basically, it was a very versatile place
to end up with things. Yeah, but again, I don't, I'd have to pull up
the paper to just, if I were in the throes of researching the, for this

(23:34):
final talk, I would have the answer for you, just like, burp. But
they, they did do a lot of work and research. And I'm saying the number
as if it's ridiculous, but it's, it's only ridiculous for
the thing we're talking about. It's not ridiculous if you're thinking
about. I want to provide constant level media
for people as a distributor of Mediaev,

(23:54):
like, it makes sense what they're doing. They did a lot of research to figure
out what would work best for them. So I don't mean to disparage them, is
what I'm saying there. So you've mentioned the talk a
couple of times that you're going to do in November. I'm just out of my
curiosity. You must be really
into just the technical, the research.

(24:16):
What's your original background? Did you come
from an electrical? I mean, obviously you're a coder, but. No, I
came from. I was a composer. Oh, okay. Yeah,
I studied music at UCSB. That's in Santa Barbara
in California. Curtis Rhodes was there at the time. He was the guy that.
I think he's like the father of granular synthesis, right? Yeah, I think sounds

(24:39):
right. But I was an undergrad so I only got to take a couple classes
with him. But I looked forward when I was at the beginning of college, I
was like, I can't wait. But I couldn't figure out how to
make my way in music after school. And so
over many years of working in kind of like embedded systems industries and other random
stuff, I taught myself how to code and

(25:00):
then eventually I had a good mind for it. I made a lot of
friends in these industries that were like really technical people.
And so then I married the two
interests. I don't know why it took me so long to do it. It took
me until I was like 40. And then I, I'm never looking back. That's awesome.
No, I love, I can tell that you're super passionate

(25:21):
about what you're doing just from like how.
Because I watched a few other things. You had a really great interview on a
show that I can't remember that I really enjoyed. Maybe
help me, Devon. It was help me to Vaughn. Yeah, yeah, yeah. Those guys are
so great. They put on a fun event
in the. Yeah. So I was just curious

(25:44):
like where that spark came from. But let's move because
we talked beforehand and there's a bunch of things that we want to get into.
We were emailing back and forth and you mentioned
that you see a place for AI and machine learning and
music. I totally agree. I see a place for it. I see
places it shouldn't go to you.

(26:06):
What is the kind of ethical lens
developers and idea people should be looking through when
you're making tools for audio and music?
I got a couple, I got a couple things. I'm going to get the ethical
thing like 1 second. But I got to go back to that split about the
technical versus the creative and the inspirational.

(26:27):
And I think that the big hype problem with AI right now is it's
conflating them. I have not seen
a mid journey dolly, whatever picture that I'm
like, oh yeah, an original thought really
makes me think right it's all, by its very
nature, derivative in a way that's inhuman. True,

(26:51):
and we are, ourselves good with patterns and stuff like that. But there's something
about human spirit and inspiration and creativity
that is very personal. And I
like the term machine learning more than AI, because it isn't an intelligence. It has
no emotions. It's not thinking about you or your family or your pets. It's
not ever going to make a hard decision in its life. And

(27:13):
looking to it to generate the art that we use to inspire ourselves in all
sorts of ways is preposterous. It's, in fact, one of the things
that makes me feel upset about
the way that people in Silicon Valley might frame their
technologies. I don't, I don't, I don't like that
feeling. And so that's, that's a

(27:34):
personal thought on my part. I think a lot of people might disagree with me
on that, but I'm, I'm standing by that statement for the rest of my life.
I don't think I'll, I'm not going to change my mind on that, so don't
try to convince me. However. Yeah, I don't know what else to say
about that. But that's just, that's what it is, right? And I think that, you
know, three or four years down the road, we'll know a little more about what
can and can't be done with certain types of stick. But the technological end, that's

(27:58):
different. Machine learning is extraordinarily
phenomenal. Mathematical pattern matching machine.
Yeah. And it does some crazy types of stuff
where we don't exactly tell it what to do, but we show it a bunch
of pieces and it kind of, you give it a process that has
all of these things that can be tuned, and they're called weights

(28:20):
eventually. And so it comes up with like, a bunch of these weights, which just
looks like a big list of floating point numbers that are like
15, 30, 30 decimal points of
whatever, because that's what it is. You're running it through some sort of crazy matrix
multiplier. It's bl and it's just sort of pattern
matching. So let's

(28:41):
say I am trying to model
an analog circuit. This is not something that I tend to
do all the time. Classically, I'd have to
say, hey, I'm going to look at the way a mathematician might
think about the parts of the circuit, and then I'm going to look at the
way that changes as you push voltages through it, and I'm going to build myself

(29:04):
a mental model of how this machine reacts to electricity flowing through
it. And I'm going to come up with math that a computer might actually be
able to do in real time. And then I'm going to tune all the
little bits and pieces so that they match what the hardware kind of does.
And you can imagine without knowing how to do any of it, that there's probably
three or four people that have figured out different techniques of doing this kind of

(29:25):
modeling and whatever, and that's what makes its way into plugins, right?
It's like, okay, well, I want to make something sound like a Moog ladder
filter. Okay, well, what are the parts of a ladder filter? What are their resistances?
What is blah, blah, blah? And then you do all these things. I'm not down
talking that, I'm just saying, like, it's a lot of stuff. I can't even begin
to explain all that stuff

(29:47):
now. And we might call that, well, I'm not going to use, you know what,
there's some terms that were called black box and white box. We're trying to move
away from them, but I don't know what the new terms are, so I don't
know what to call them. But essentially I'll call it a transparent box and an
opaque box that would be a transparent box, because
you can see all the pieces you've tried to put together and all the stuff
you're trying to do and da da da da da. Right? Machine learning gives us

(30:07):
an approach where like, you know what? I'll just throw a bunch of random signals
into this, into this algorithm, and it's going to look at the input
and the output, and it's going to just sort of change all
these little things around in the weights that I've set up for it, the neurons,
and I'll come up with an approximation that when you
put stuff in, it generally looks like what you told me comes out of it,

(30:29):
but you have no understanding of why. Right, right.
So it's really interesting is that if there is
a piece of gear or a feeling or thought or whatever that you're trying to
model and you don't know how to model it, or maybe it'd be too mathematically
complex or complicated for a machine to do in real time. AI is amazing
for that. I don't know whether it's better or worse for things that we have,

(30:50):
like models for and stuff like that, but it's an amazing tool
for. We don't have the mathematical
theoretical lens to look at this through.
So can a computer figure it out for us,
maybe. Yeah, in some cases, yes, it definitely can. And
what's even cooler is that I'm working with

(31:11):
this other guy whose name is Jotun Chowdhury. He's done some really cool
stuff, and he told me one day that one of the other interesting things about
it is sometimes you learn from the models because you're throwing all this different data
at it, what some important things are for the
input. Oh, I had no idea that in order to reach this
goal, it's important to know this thing about the audio,

(31:32):
which might lead you to get a better understanding of what's going on under the
hood, and then you might not even use the model or AI in the long
run. It may have simply pointed you in the right direction so you could do
something else. So there's all sorts of ways. Anyway, anytime you're trying to
do some sort of pattern matching or whatever, I think it's really cool. I'm
not a huge fan of generative AI for the reasons I spoke about before. I
think that it's, it's really good at giving you grey goo.

(31:56):
If what you want is gray goo, go get it.
There's my alliteration for the day. Go get your grey goo. Yes.
Maybe don't put that in because people could misinterpret it. But that's kind of just
the nanobot thing. Right. But for gray matter, like, in your brain. I don't know.
Right. You get it. No, that's. I never actually thought about
machine. Cause I think, like, use case, like, where. Yeah, where can

(32:18):
machine learning or AI speed up? Like, my, my workflow,
but, but make me more creative. But I actually didn't think about the
software side, like, finding a more accurate model or a
more, uh, a better, a better plugin. Not that I
want another 1176 plugin, but. Exactly. Yeah, we don't, we don't need that.
But we. There is gear we might not be able to measure, or it might

(32:39):
be able to do it really efficiently down the road if we find out there's
a special way to do it. Yeah. Oh, but the other thing you said, too,
is also really important, and it's kind of the same thing, though. It's like,
in a weird way, it's like, it's very good at pattern matching, so it should
be able to get rid of minutiae. Minutia in
our lives. Right? Yeah. For mathematicians, that might mean modeling something. That's

(33:00):
just like, I don't want to go through the motions of modeling this thing. I
know how to do it. It'll take me a week, but I'll just throw it.
Right? Yeah, because it wouldn't make a difference. Like, if it wouldn't make a difference,
why not use it? Right? Right. So anything that's a productivity tool or
where we're going through the motions non creatively. Right.
Level matching, ducking things. I

(33:21):
have a cousin who works in video editing. He was talking about this today. There's
a lot of just technical phases or
phases of production you go through where you just have to go through the motions
of doing this really boring thing. And tools will get better and better at doing
those boring things for you. Yeah, it's impressive. I
mean, I don't really know any code, but I'll

(33:43):
build a soundflow script to do something, or I'll take one that
I have. I don't know if you're familiar with soundflow, throw it into GPT
and be like, hey, this is what this does. This is what I want it
to do. And it spits out code and I paste it in. It works. I'm
like, this is sick. That stuff is great. I'll never learn how to code now
because now I have the box. That's it. There
you go. It's another pattern matching, transparent box.

(34:07):
Yeah, right. Where you're like. Or, no, no, sorry. Pattern matching, opaque box. My
bad. Where you're just like, I know what I need to do. I know it
can do it. I know the input. I do the thing and it does the
thing. It does the thing. And that's all great. So to the degree
in which you can eliminate those things, that's great. If you want to
make a movie and you don't want to pay a composer, and instead you want
bland music that's incredibly derivative and makes weird decisions that

(34:30):
then you can use generative AI. But that's where I'm like, I don't like that
ethical boundary that much. But, like, still, okay, if you want to
make poor, mediocre stuff, like, go for it,
that doesn't mean you can't use generative AI as part of a process.
I mean, you can do whatever you want, but I'm sure there's lots of
amazing creative ways we could do that. But the emphasis right now seems to

(34:53):
be on mimicry, which I don't love. No, I
think it's the flashiest, right? You're talking to somebody that
doesn't make music, and you show them that the computer made
a song. Like, the musician's not impressed, but
that's only 0.05% of the world. Everybody else is
like, whoa, that's mind blowing. So I feel like it's the flash

(35:16):
is why the generative mimic is the thing. Because
it's bitter for the share price. It is. It is
bitter. And I don't mean to say if you take, like,
backyard potluck video of you and your friends and you don't know anything about making
music, you're like, someone make music for me and it'll be needing to the video
and you get to do it, that there's something that horrible about it, like, no,

(35:38):
actually, that's fine. Not everyone knows someone who can do that
kind of thing. And you're not going to pay licensing fees for your own
video. So stuff like Suno AI is cool for that. But the ethical
barrier there is what aren't they telling us about the training data?
That's something that I believe the artist should have. Right.
Just to. When it's being used for that kind of thing, we just don't know.

(35:59):
It's very. Again, it's an opaque box. Yeah. Yeah. That's
something that I think is going to be really interesting over the next couple of
years, is the training data. And I think that's going to stretch across all
the AI because I think people are starting to realize
that there's value in licensing your voice or value
in licensing your footage or whatever, but if it's just being taken

(36:20):
off and used, then you're missing that opportunity.
Right. I did see that this is kind of old news, but
YouTube was doing a thing for shorts. I don't know if it ever launched,
but they had five or six voices. Charlie
Puth was one of them. I mean, all names, you know, and you were gonna
be able to generate like a 32nd, 45 2nd piece of custom music

(36:42):
for your YouTube short. And I don't know if that
ever came to light, but there's obviously business opportunities for people that are
willing to sell their thing. Yeah. And
I think it would also be really interesting to be able to train an AI
model with your own music and be like, what would my, you know, weird
thing, like inspiration or something, you know, there's cases for it.

(37:04):
Yeah, they're interesting. There's interesting cases for it. Whether they would
work or whether they would, you know, there's some stuff out. There,
if you don't mind me, bring it back to the mastering conversation, though. It's all
back to that kind of spark of, you want something that's going to satisfy a
requirement for you, or do you want something that is going to inspire you. Yeah.
And I think if you want to satisfy a requirement, there are applications for machine

(37:25):
learning that are pretty, they're pretty good, but if you want something that's going to
inspire you, which I think is something most people actually want out of their
music. That's why I think AI doesn't really have a great place
in, um, in. In that part of the. The phase, because
it's. It's saying, I'm going to make these
creative decisions without a soul. And you're okay with that? So.

(37:45):
Okay. Yeah. Do you. This is kind of related to pattern matching, but
back, back to mastering. Do you ever feel like the new
play, like the gullfoss, the soothe, or just
isotope tonal balance? And I'm not. I use all these tools. I I'm not. I'm
not railing on these companies. Yeah, yeah. The fact that these things are matching
some kind of algorithm curve or making suggestions based on what's

(38:08):
pleasing to the ear. Do you think that's hurting
young engineers who are. Who should be shaping their taste and, like, maybe
your taste is dirty and bright, but you're never going to find it because you
put soothe on everything? I don't know the answer to
that. I don't know.
I want to say no. I want to say no. It's not. I think that

(38:30):
that's one of the things where it's like, any tool can be
misapplied and it doesn't make the tool necessarily bad.
There are some tools that are necessarily bad, but I would
not. The tools that you mentioned, I think they have legitimate uses. I love them.
And I think that for people that are listening for something in specific, they offer
a great way to get there without doing a bunch of technical

(38:53):
work. And that's the thing that I advocate for. Let's get rid of the technical
hurdles so you can spend a lot more time on the creative hurdles. That's what
we want to do as musicians, as producers, as mixers. That's the part
that gets us going. We all. I mean, I love technical stuff, too. Right. But
I know when I'm in the creation mode, like, like
if I'm just geeking out, no one's gonna want to hear it. There things like

(39:13):
make an extra difference. And when you're making music and if you're. Especially if you're
talking about, I don't know, like youth, kids,
whatever, whatever that is now, they're taking the most risks of
anybody. Young people tend to take the most risks. They're trying to
stand out. So if there's a tool, they're also gonna be the ones that are
most likely to either totally reject it or totally abuse it in a way that's

(39:34):
totally unique. So I think the more tools that exist for that thing, that's great.
Yeah, I agree with that. Yeah, there's. I'm not a
curmudgeon. I definitely, as I get older,
I often, like, wonder when files come into mix. I'm like, what the fuck did
you do? How'd you make it sound like this? This is
cool. Yeah, they don't know the rules. Can I ask, or is it not

(39:57):
polite to ask because I know you did it?
Weird. But, yeah, no, there is.
That's why it's nice to collaborate with people of, like, all ages and, like, step
outside your circle, because you're like, oh, shit, my friends, don't do. Don't do
that. I like that trick. You know, the human aspect is. I don't
know, it's just the thing that is never going to go away. Right. Sitting in

(40:19):
a room with people making music. That's why we all started this,
right? It's how we're going to keep doing it. That's right.
Yeah. Before we go, I love people that
build something. Right. Whether that's a product like you have or if it's. Maybe
it's just like a unique career perspective or
direction. Right. And so I think there's a lot to take away from those stories.

(40:41):
So, a, can you share what the core mission behind
starting music hack was? And b,
are there any takeaways from the journey so far of building a
business that you think I are valuable to a producer engineer's
career? Oh, wow.
Yeah, it's really been a journey. So Stan and I talked

(41:02):
about this for a while. We met through a mutual friend, and then we talked
about the mastering chain for a while. And the idea really
was to get rid of those technical hurdles, to get rid of.
To make people feel like they can be confident that their taste,
that they know what they want, and that that's enough. You don't have to have
some kind of deep wizardry or technical

(41:25):
knowledge or insane amount of control in
order to make something properly mastered sounding great. If you have
a good mix coming in, you can have a fantastic master track coming
out, and you can spend your time on the creative aspects of it
rather than wondering what the technical trick is that would get you
where you wanted to be. And that's why everything on master plan is named, like,

(41:47):
loud, wide, high, thick. It's named after the
quality that you want to change in the sound. And I want to make tools
like that. Yeah. Because you don't make instruments like
a violin doesn't have, like, 500 knobs on it, that each have a different
name. It's just like you got them strings, you know
what I mean? Yeah. And you just gotta move them. Right. Yeah.

(42:08):
And I think that making something that is eminently
playable and musical
and that gets rid of this sort of gatekeeping
attitude of technical excellence
and precision as gospel, we
wanted to make something that anyone could feel confident taking their taste and

(42:30):
applying it to a track. Yeah, that was it.
Actually, while you were describing that, I was like, oh, they really put this into
the plugin because I was thinking about the way that it's named and
how I've been using it is I don't really care what
the EQ frequency is or what the tape button
does and what happens on the right side or the left side. I just press

(42:53):
the button and I move it, and I'm like, let's go this way. Yeah, that's
better. And it's quick to just shape, and I feel
like you guys really epitomize that statement.
So congrats. Yeah. As far
as the journey, I think the most applicable things are
you have to be determined and persistent. It doesn't matter what you're

(43:15):
doing. You have to do it a lot, and you have to. When you get
thrown a curveball or when something doesn't go your way or when something you thought
would work just falls flat on it, you just got to try something else or
try it again. Keep pushing for what you know
is what you want to create, and you will find your way there. Because it's
never easy to get something amazing that you love. If you

(43:36):
are a self critical person, and then the world is going to
criticize you, too, when you put it out there and you have to think about
what are all the things that went into. You just got to keep
going. Some things are ridiculous, and some things are frustrating, and some things
are confusing, but all of them will respond to you trying
again, trying a little bit differently, putting yourself out there again.

(44:00):
I say this as somebody who personally stopped doing production,
but I didn't. But I stopped doing production because when I spent
time in some camps and stuff like that, I found that, you know, what
I'm most myself and best put to use when I'm helping
people with the tools that they're using to do the production, I'm more
effective at, like, being part of the group when

(44:21):
that's what I'm doing, so that's what I'm going to do. And I can feel
good about myself for that. So the other thing that I think that's worth doing
is whatever you're part of, try
to find what is it that you feel the most comfortable in that role, and
own that role. It might not be the role you want. The joke I think
I told on help me to honor something else. It's not even a joke. It's.
The truth is, like, I just. I thought I was going to make some sick

(44:44):
synth rifts, but guess what? They weren't sick enough. So that's. And you
know what? That's fine. I found something that I could be proud of. That was
something else. That's cool. Yeah. And it's not
always gonna be what you want, but if you can take pride in it
and. And own it, like, it can be really fun. Yeah. And you
have to find the people around you that understand that's what you're doing, that they.

(45:05):
They get that all that is stuff that transfers to anything you're doing, even
outside of music. I don't care. Everything. Everything there was. While you were saying that,
there's a. There was a phrase. There's a phrase that came. Came to mind, and
I feel like this phrase has kind of. Kind of been
twisted around the wrong way. That phrase of, like, you know,
failing your way to the top. And I think people, people use that in a

(45:26):
negative connotation now. Yeah, but it's
all about the mistakes of learning. It's like any entrepreneur, like, and you
will tell you how many times they failed at something and what
they took away from it. And I kind of hope that that phrase started there,
and now it's been, like, twisted around, but I don't know why that came to
me. You're just talking about, like, growth and making mistakes. And I'm like, no, that's

(45:48):
the thing. We're using this phrase wrong.
Yeah. It's like failing. Failing up isn't like, oh, you
got promoted, even though you suck, or like, some pay attention, though. Even you suck.
And it doesn't mean faking it when you don't really have
anything. It means making an earnest attempt at doing
what you really want to do and making an earnest attempt at listening

(46:09):
when people receive it and finding out what it is works and
doesn't it? Yeah. And that entails all kinds of emotions. I
can't tell you what you're going to go through, but it's going to be something.
Everything you just said, really fits with the core
of the show. Your career is going to go in different directions. I
would never believe that I would be doing a podcast. And as soon as

(46:31):
you fixate your success on the role that you want, as soon as
that role isn't exactly what you thought, it's a
big letdown. It's like you have to understand that things change,
and when you find your place, it's gonna be the right place, you know?
But, yeah, that is great insight. People should go back and listen to it
again. Yeah. Oh, yeah, I gotta listen to it on a bad day.

(46:53):
Yeah, exactly. That's the other thing. It's like I gotta remind yourself.
Yeah. Yeah, we
all have those days, and then it's really hard. And the trick is to
just try to bounce. Try to bounce back. I got it. Yeah.
I'll see if I can find that clip, and I'll just try to inspire myself
on a random Tuesday. I don't know. I'll clip it out. I'll send it to
you. Yeah. Hey, thanks. I'll tell my editor to find it, and I'll

(47:17):
make it minus six lufts, but it'll sound good.
Oh, no. So I've got two questions I ask everybody before we go.
One is, has there ever been a time in your career where you chose
to redefine what success meant? When you say career, I'm thinking,
broadly speaking, life since kneading. Yeah. That would be
when I left my job working for someone else to do this. I

(47:39):
changed from trying to convince other people that I worked with
that I knew what would make a good
product, or I knew how to help people with whatever the
tool was that we were making. And I realized you
can't. You're really lucky if you get a chance to do that. You have to
be at a very small company or know someone really in a special way. But

(48:01):
if that's what you want, you can't do it.
Maybe that's a cynical look, but it's really, really, really, really hard to get
people to deploy resources on your behalf. That's true. That's
true. I agree. For me, my goal changed when I
realized I can't be doing that because I want to make
something whole from nothing, which, it's okay if you don't want to

(48:23):
do that. Some people, it's like, I want to
focus on making this one thing that I know about this process better, and it's
much easier to find your right place in that. So I don't want to
be disparaging to people like, that's not what I'm trying to say here, but that's
when my goal changed from please this other person so that
I can get what I want out to just make the thing,

(48:46):
get the people around you who will help you make the thing and make the
thing. And I
make it so that people can, you know, make some
tool that people really can use to make some stuff that they love. Yeah. As
opposed to, like, help some company make some tool that they can use to
sell to a defense contractor. So, yeah, my goal

(49:09):
changed. My goal changed there. Yeah. Fair, fair.
I love that. I have a lot of parallels in my life. I won't go
into it, but, yeah, yeah, but yeah, there is. When you want to
do your thing and you're in a place where you can't do your thing,
it's so frustrating. So I feel you.
Feel you. Drives you crazy. Yeah. But last question

(49:30):
before we go is what is your current biggest goal that you can share with
us and what's the next smallest step you're going to take to go towards that
goal? So right now we have a plugin
that, well, we feel, but I think a lot of people that are using it
feel like you can put it on a whole. On the two track, on a
whole production, and almost anything will

(49:51):
benefit from it. The people that start using it just, they know they can put
it on anything. My next goal is to make something
for that, that has that on an individual track
basis where you could put it on any track and it has a real lot
of useful stuff you could do with it because that's a whole different
animal than a mixed bus or track. And I want to make it

(50:11):
template worthy. Just like, I feel like master plan is where people just. It's
a go to. Yeah, I know that doesn't sound like a lofty goal, but it
is like making a template worthy plugin is really hard. Making one for the two
bus or for a mix bus versus one for individual tracks is even
harder. And so
that's. Yeah, that's what that is.

(50:33):
I don't know if that's. I wish I could say something more lofty, but it's
got to have all of those things. It has to be indispensable, it has to
be very musical and creative, and it has to be
uncomplicated but very expressive. Like these are. These are lofty
goals. And the smallest thing I'm doing to this aim is actually very
technical. And that's that I myself make myself
tools so that it's easier to do what I'm doing.

(50:57):
And so I'm changing the whole way that we're doing
graphics and stuff. And so I just want to finish
that, which, relatively, is a small step. But once
it's done, I can put it to rest and focus just on
making more of the best plugins I can make.
Yeah. And it's almost silly at this point because it's like,

(51:18):
man, I've wanted to release this thing for so long because we could have been
this one plugin company. I'm like, we got it. We already. We have the other
one. Can't say what it is yet, but
this has been holding me back, and I just can't wait to get it behind
me. So that's the small step I'm taking, and it will
open up a lot of ui ux
things that I think will be very, just delightful stuff that'll be,

(51:41):
like, make it more pleasing to use. That's awesome.
Well, I'll look out for it. Do you think it's a far way
away or a close. No. Okay. It's a close way away. Like,
because, you know, there's, like, who knows? There's all of the little
things you have to do that you don't expect. But, like, yeah, I want it
out of nice. That's awesome. If other people want it out, believe me, no

(52:05):
one wants it more than me. Maybe. Maybe Stan and Alex,
the other two music hackers do, but, yeah,
that's great. That's great. Well, Sam, please tell people where they could find you, where
they can find music hack, where they can maybe see your thing
later from the developers conference. Whatever you want to share, go for it.
Yeah, you can find me. I'm actually on a discord with a

(52:27):
music hack discord, which you can find at our website,
www.musichack.com.
and you can also just look up master planning. There's, like, videos of a bunch
of stuff that. Help me, Devon. If you just look up my name. Help me,
Devon. Sam Fishman. Maybe he'll spell it out in this
thing or like, that. They'll. They'll find it there. The conference. They
take a while. Cause, you know, like, they take registration fees for people to listen,

(52:50):
and then, like a year, like, oh, yeah, seven, eight months later, they released the
talk. But I did a talk about oversampling. Okay. Last year.
So if you're interested in that topic, you can look up ADC.
Oversampling. And my name will probably pop up there. ADC is the audio
developer conference. If you're in the audio world, and you're interested
in engaging with developers in any sort of way

(53:12):
or just curious about what's going on in our world. Registering for that conference and
checking it out is really cool. Where's it at? It's in Bristol. But you can
do it remotely, you can do it online. That costs money though. But there's
diversity scholarship, stuff like that. I don't mean to be an ad for them, but
you know, if you're, if you want to meet developers or you're like a producer
mixer or something and you're looking for someone to collaborate with, it's a great spot

(53:32):
to be. Those are the spots I hang out. My instagram is Sam
makesfood. Love it because I cook a lot. That's my other
thing. And our music hack, one is music hack underscore.
So yeah, if you want to reach out, reach out. I'm around.
Yeah, I will. I'll put a lot of that. Links to most of that in
the notes. Cool. Yeah, this is a lot of fun. I'm glad we got to

(53:53):
connect one more thing. Yeah. I'm also totally love
technical deep dive. So if like you're seeing this and you're like, I really want
to know how something works or whatever, if I can answer the question, like
I will do. If I can answer in 90 seconds, I'll just do a video
of it and throw it up on Instagram. So like ask me questions on the
music hack Instagram if you have them. I'm game. Love it. Love it.

(54:14):
Yeah. All right, somebody get nerdy and send him a
message because I want to. I want to see it go down. Awesome, man. Well,
thank you. Are you in LA? I am. I'm in downtown LA. Oh, okay. I'm
in Pasadena. I'm gonna send you an email. We'll get a beer someday.
Yeah, let's make it happen. That's it for this week's episode of Progressions. Thank you
so much for watching or listening. Be sure to check out all the links and

(54:36):
resources mentioned in the episode down below, in the video description or in
your podcast show notes. If you're listening to this as an audio
podcast, please leave a review on Apple or Spotify. It helps the show so
much. And if you're watching on YouTube, feel free to drop any thoughts or
questions about the episode down below. Let's keep the conversation going.
For those of you watching, you'll be getting a link to another episode you might

(54:57):
enjoy popping up somewhere right about now. And for those of you
listening, check out the YouTube hit that subscribe button if you haven't
already. And I will see y'all next time.

All Episodes

Sam Fischmann - The TRUTH About Mastering in Modern Music Production (AKA: It's OK To Master Your Own Music)

Episode Transcript

Popular Podcasts

Stuff You Should Know

Dateline NBC

On Purpose with Jay Shetty

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}Sam Fischmann - The TRUTH About Mastering in Modern Music Production (AKA: It's OK To Master Your Own Music)