The Illiac Suite - Music and AI

This is an automated AI transcript. Please forgive the mistakes!

This is the Iliac Suite, a podcast on AI -driven music. Join me as we dive into
the ever -evolving world of AI -generated music, where algorithms become the composers
and machines become the virtuosos. Yes, this music and text was written by a
computer, and I am not real, but... I am and my name is Dennis Kastrup.
Hello, humans. This is the first episode of the Iliac Suite, a podcast about music
made by artificial intelligence or better made with the help of artificial
intelligence and we start today with a song called "U -U -H" Written,
performed and produced by the band The Schizonoid. Although that is not really true,
the band does not exist. All is written by an AI and who actually is behind this
I will tell you after the song here on the Iliac Suite named after the first music
piece a computer wrote in 1957
(upbeat music)
The schizonoid with "u". Before I will tell you the story behind them a few words
in general. AI is at the moment everywhere. It is the hype, it is the topic and
I'm aware that not all of what happens with it evokes optimistic vibes. I totally
totally understand. We have already had many warnings from countless scientists,
entrepreneurs or programmers about the impact that artificial intelligence will have on
us and our society. That is very complex and also very scary and I'm also really
scared where this will lead us in the end. But we have to live with it. Now the
box of the Pandora is open and we have to find the positive sides of it too. So
let's focus on the music. Yes, we should consider also the danger of tech monopolies
that will control the music markets more and more with their AIs. Also,
yes, musicians might lose their income. Yes, copyrights, questions are really
important. I mean, where do we start? Data sets use millions of songs,
the copyrights are not clear at all. But other jobs will come up. Art will never
be the same from this point on. There will be a new wave of creativity we have
not imagined yet. We can just let the music speak by itself. So you will hear on
the Iliac Suite some of that AI music. I'm aware by playing these songs, I am
currently in a gray zone considering copyrights. But let's see where and how far can
we go. For example, let's talk about the song "U" from Schizonoid. We just heard.
What does the web page say about them? The Schizonoid are a truly unique band that
have taken the music world by storm with their innovative use of technology and
their robotic band members. The band was formed with the goal of pushing the
boundaries of what is possible in music and they have certainly succeeded in doing
so. They are on the first music label signing only artificial intelligence artists
called Operae. I talked with the persons behind the label about their vision. But
before that, let's hear another song by the schizonoid puppy's theme.
Schisanoid with Pappi's theme. Let's get into what all this is about. The band is
on the label Operae Records Opera AI run by two Italians Remo De Vico and Jacopo
Solari. Remo is an experimental music composer active in theater cinema and
performance art based in Italy. Jacopo is a physicist who works with AI.
He is a guitarist and music producer based in Berlin right now. They have been
close friends from their university time and Operai was founded in February 2023.
How that happened, Jacopo told me in an interview earlier this year. It was the
period of when Chatsy PT came out last November and proved to the world that the
AI revolution is at our doors and we've always been brainstorming or talking about
new ways of making music, the interception between music and technology and yeah I
remember this phone call where Remo called me and introduced this idea and it
escalated very quickly because it was, "Hi, we could do an AI, a label where the
artists are AIs." And we were like, "Yeah, cool. And then we could do that. And
then we could do that." And, you know, it escalated very quickly. And we got more
and more excited as the more ideas we're putting on the table. But essentially,
we wanted to make an art project. We don't want to have a major label like Sony.
We like, we are both researchers, I was a researcher in science, he was a research
in music, so we like working at the edge of the unknown and this new technology
that was out of course for a few years already, but chat GPT really showed that
you can interact with this technology, with this knowledge in many different ways,
many novel ways. The possibilities and the creative territory was very fascinating for
us. So we jumped into it without a clear business plan.
We are not businessmen.
But yeah, we just wanted to have fun, start experimenting with this idea and create
the first, let's say Our plan was to have a catalogue, to create a big catalogue
as soon as possible. And indeed the name comes also from that in a way, because
Opera AI, in addition to be a similar sound wise to open AI,
also means opera AI means workers, like worker class workers in Italian.
So we wanted the AI to be our worker, to make our work as a musician less
tedious, more direct and see what creative possibility would open if we were lifted
from repetitive work, for instance, instead of being there and say, "Okay, now I
need to come up with a melody or a sound scape for my next track. Let's start
with a let's see what what eat. Let's just give it an "it" pronoun.
Sometimes I hear you say "he /she". But let's see what it gives to us and let's
start from there. So that if we have many different foundations, just at our
fingertips, because you can ask these models to give you any kind of outputs and it
gives you instantaneously, we could build very fast a lot of trucks and make this
label happen. So let's hear what they did so far. This is Sushi Time by the artist
called Kawahara Yosai.
Kawahara Yosai with Sushi Time. So who is this guy? He has dark hair and a
moustache and comes from Japan. The official information goes like this. Kawahara
Yosai is a renowned composer and multi -instrumentalist who has made a significant
impact on the Japanese music scene with his unique blend of pop, experimental and
traditional sounds. Born and raised in a small rural town in Japan, Yosai was
surrounded by the rich culture and history of his country from an early age. But
how exactly do Remo and Jacopo write his music? Let's do chronologically because it
makes more sense. We started with the idea of using Aiva, which is this you know
about it, which is quite the state of the art I would say for complete composition
generation via AI. So with that you can create what is called a generation profile,
which requires a lot of different inputs, like the course of tempo, what type of
chords, what type of rhythms, and you can choose among some pre -made by them,
by the company, IVA, patterns or feeling, and then you can generate a lot of
compositions via the same generation profile, and the AI somehow randomizes these
parameters and outputs, let's say 10 different tracks from the same generation
profile. So they're on the same style. And I have to say, most of them at the
beginning, they were horrible, they didn't work. And that's improved as we learn how
to use the program more, but at the end of the day, you still have to generate
maybe 20 or even 50 composition before you found that two or three that you really
say, "Oh, wow, this is cool. I want to work on it." And then what we did at the
beginning, we took the output from IVA, which is basically a music score written for
computers via MIDI, this protocol that electronic musicians and producer use, and we
would curate the sound of the composition. We would keep all the section as the AI
output outputted it, maybe some adjustment here and there, and we would substitute
the sound that we like using our own hardware synthesizers or our software
synthesizers inside programs like Ableton or Logic, like what is called a DAW,
digital audio workstation. It allows you to allow DAWs to take the output of IVA
and arrange it sound -wise as we like. So in this way,
we call ourselves "curator" of the artist. And this also allows me to say that,
in our opinion, the human intervention is still essential in order to compose with
AI. We don't make AI -composed music for us, we try to interact with it.
We hear another song from the Japanese composer Kawahara Yosai "Women of Ireland" and
I think it really works. You can feel Ireland while listening to it, but make up
your own mind, women of Ireland.
"Women of Highland" from Kawahara Yosai. I do not know in which way this song was
actually written, but Jacopo tells us now the other way of writing the songs with
the help of AI. We start finding many other different ways. Remo started asking
Chopped GPT to write code for Sonic PI, which is a programming environment for
sound, or help him giving an idea for Max MSP touches,
and that would cut down the time that you have to put, click, and find an
experiment. You get something that already is working, and personally, I believe that
especially in electronic music, when the possibilities are endless, if you find a way
to limit yourself, say, "I only use this synthesizer for this track. I only use
this instead of having all the time, the full pallets are sound available, it's very
helpful for productivity. And AI, one of the good things that we found is that
limits yourself. It gives you something that it's already kind of working. And then
you can work on top of it. And it's a choice. It doesn't mean that it's better
than what is before. And we don't even know if it's gonna be better, but we are
pioneering all these different ways. Another, Another, so both me and Ramo found
different styles of interacting with the AI. I told a couple of the ones that Ramo
uses. Last time I asked ChatGPD to write the arrangement for me for a song in the
style of Daft Punk, let's say. And then I started asking more in detail the output
that I wanted, because you need to ask the right question to the AI in order to
get the right answers. And I ended up with time stamps, say from 0 seconds to 30
seconds, from 1 minute to 2 minutes, and descriptions of this time interval of what
was going on. So the drum is coming, now the bass is leaving, now there is a
melody, and then I asked again, "Oh, write me the melody for the section that you
mentioned." And then the melody came out. Then I took some old output from Aiva
that I never used because I didn't like it, but I took the MIDI was there in this
project that I'm using. I tried it, it worked. So it's not always like one single
workflow. You can mix them, you can experiment with different models.
There is text to speech, there is text to music, there is so many different things
and it's so fun to interact with it.
We just heard Helmut Cartesius with "Bear Fronts", the webpage says about him.
Helmut Cartesius is a visionary composer and sound designer whose work seamlessly
blends elements of ambient, classical and experimental music. Born in the 1970s,
he began his musical journey as a pianist and quickly fell in love with the
possibilities of sound manipulation using magnetic tapes. - But where do these
characters like Helmut Cartesius and Kava Ara Yosai actually come from?
- Those are fictional characters. Those were supposed to be the AI entities that I
mentioned earlier. But of course, the AI is not so advanced to be a person.
So we are the curator, we are the ghost writer of these fictional characters. But
all the material that you see online is generated by AI. We do the profile pictures
with Dalí. So those faces that you see, they're not real people. It's just Dalí
making characters for us based on our prompt. And for instance, I'm curating Tramco,
which is a techno DJ from Berlin. So I asked the chat GPT to write a biography
for it. and I gave it some more details about how I wanted my character,
my alter ego, my AI avatar to look like or to sound like.
"Tremco with telecom revolution, so who is this guy, Tremco?" Tremco,
short for Tim Remco, is a techno producer from Berlin, Germany, who has made a name
for himself in the electronic music scene with his unique blend of hard -hitting
beats and melodic soundscapes. Here he goes again with Silicon Heart.
Tremco from the label Opera. As with many artists in the field of AI music,
Remo and Jacopo are aware of the fact that AI will shake the music industry.
Again, like so many electronic devices did before. They both are musicians by heart,
but they also believe we should embrace the change. Out there there's a lot of AI
skeptical in the music making community and it's totally understandable. And on
Reddit, I think I posted by mistake with my personal profile, but I made a post on
Futurology, a very big Reddit and the post got 40 ,000 views in a few hours,
but it also got a lot of negative comments. People saying, "As if musicians were
not struggling enough already, or this is very dystopic, AI music sucks,
you know, and we were expecting it. And as let's say,
we're also not political, but let me cite this Italian character is not everybody.
The Lusconi taught us that it doesn't matter if they talk badly or good about you.
The important thing is that people talk about you. So in that sense was a part of
our plan, but yeah, we totally understand the worries in the music community, and we
agree with the fact that AI music, as it is without music intervention, sucks.
But we try to see it from a different angle. We just forget about all the morals
and all the ethical worries that, you know,
this revolution is carrying and it's producing. It does look like Italy that banned
chat GPT, you cannot access it, it's from Italy. But yeah,
we just try to have fun and to imagine how it will work.
And with these words, we are ending the first episode of the Iliac Suite. I hope
you enjoyed it.
I'm excited to share new AI music and new artists with you in the next episodes.
Any comments? Write to me, mail @deniscastrup .com. You find that address also in the
notes of this episode. Please also write me if you make AI music I'm curious to
hear more and know more about it and maybe play it here in the podcast So this
was it first episode take care humans and behave
(upbeat music)