The Illiac Suite - Music and AI

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

"Hello, humans. I hope you had a great summer and you're enjoying the last summer
days if you are listening from the northern hemisphere. It was a good summer for
me, I have to say, but I realized something. Even artificial intelligences are on
vacation. I had the feeling in the last months I have not heard about many AIs
that are Russianizing the industry anymore. By the way,
what a bad description or bad term because for me, a revolution is something totally
different than AI companies branding their product with a revolution so they can make
more money. Well, okay, there has been the answers of Udyu and Sunoo who responded
to the copyright claim of music labels some weeks ago. So something happened there.
And of course, they argued and defended themselves with the term "fair use" in order
to support their position. What I think will happen in the future? This will go on
for some years. One party will accuse the other. Nothing will be ruled by courts
and one day they will find a solution that suits both of them. Will the artists
profit from that? I don't know. I doubt it. But what I do know,
there is a company that has just announced that it will pay artists who contribute
music to their data sets. And we will talk about that in today's episode after the
intro.
This is The Iliac Suite, a podcast on AI -driven music. Join me as we dive into
the ever -evolving world of AI and 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.
So what company am I talking about? Lander, the music creation platform based in
Montreal, which works with an artificial intelligence. Some weeks ago they announced
on their site "The Lander Fairtrade AI program offers new opportunities for musicians
to monetize their work and participate in the evolving landscape of AI in music." Of
course I wanted to know more about that because I see myself in the fight that
musicians should get paid for their contribution in datasets, which are, in my
opinion, not fair use. I reached out to Lander and got to talk to Frédéric Ranger,
the vice president of marketing and communications of Lander. Let him tell you first
a bit about the road Lander has taken until today. More than 10 years ago there
was a technology that was at Queen Mary University in London and there was a group
of people here in Montreal that saw that and said, hey, this is an opportunity to
bring more technology into the hands of musicians so they can benefit from these
great technological advancement. And it started as a platform to master your songs.
And this was something that, you know, it's always been a bit mystique. It's always
been a bit hard for musicians to do. And quite frankly, they needed to hire some
very expensive engineers back then. And so Lander said, hey, let's simplify the
process. Let's give access to this mastering tool, automated, to as much people as
we can. So they launched the Lender Mastering software, which was online back then,
and it was a huge success. And it's always been into Lender's DNA to work for
musicians. So we always tried to find ways to add to this product. And so we added
some plugins, we added some samples, we added the distribution service, and in the
end, if you look at how the platform is today, it's literally, as I said,
everything you need to make music, distribute music, and monetize music. Then, of
course, with the past, I would say two years, we've been asking ourselves a lot of
questions about AI, and by the way, back then, 10 years ago, it was already AI.
Maybe it was called machine learning with the mastering plugin, and it was trained
ethically, which was already a term that we were using back then and giving it back
to the product. But now with the recent AI for AI, I would say we said, how can
we bring that same brand DNA that we work for musicians? How can we bring more
possibilities for musicians to monetize their work? And we said, okay, let's do it
right. There's a lot of companies that are trying to go about AI right now and we
said, if there's one company that can do it right, it's Lander. Because we have the
user base that we can ask questions to, we have the technology that we can put
forward and quite frankly we have the way to make it available for users to
monetize their work. So we said let's take the same approach we did with
distribution. Let's make sure that we ask people for their opt -in so we can use
their songs. We have a lot of distribution being done through Lander. So we can use
their song to train our AI models and it would be done ethically. Again it's not
scraping the internet, it's not going on Spotify without any consent, people have to
opt -in. And then we are going to be basically what people say is,
you know, you can use my tracks to train your AI and people will get compensated
for it. So this was kind of our way to connect it back to AI today and take
basically the best of both worlds, which is making sure that people can get paid
for their work and that we do it ethically throughout the whole process.
So they have been one of the first ones in that field to adapt AI for mastering
for example. I remember also that I have done an interview with Pascal Pilon in
2018 and we already talked back then about their AI for Mastering and I also
remember that this was so weird for me to understand a computer. Does all this
work? While times are changing back then it was just called machine learning and it
is still running our Mastering plug -in. There's a plug -in version now and our
mastering engine has been trained over millions and millions of hours of you know
real humans tweaking some songs and so on and so forth and also making sure that
the machine learns from those tweaking and what sounds best. Actually, if you look
at a concrete example, right now if you upload your track to mastering service or
if you use the plugin, you'll get three ways to master your songs. You get a
balance, you get an open and you get a warm type of setting. While this, again,
reacts differently if you have a hip hop track than if you have a rock track and
so on and so forth. So it's really perfecting the art of giving you as much value
in terms of using the product and getting you to the finish line so you can
distribute your song, you can monetize your song if you want. So the AI that we
use is assistive and I want to insist on that word because it's not meant to
replace the musician or the technician for that matter or engineer for that matter,
some engineers actually use our products to get them to 90 % there and then they do
their secret sauce and then they send it back to the client. But if you're an
artist, you can use all of the suite of product of Lender without having to touch
or without having to go to a real studio. You can use some of our plugins that we
have for vocals and make them sound great. We have Syncore Arts, which is a suite
of vocal plugins. We have plugins plugins for basically everything, since and so on
and so forth, so the way the platform operates is that it guides you through the
whole process, the whole creative process. The best comment I have on the platform
is users coming back to me saying, "Fred, we use the platform and we don't even
feel that there's AI or technology that's backing the whole process up. It's a
design system that gets me from thinking and dreaming about a song to distributing
it on Spotify. So that's the best way I can explain it. Millions of input points,
millions of hours are being spent tweaking our products and so on and so forth.
Again, always with consent, there's no scraping of any data of our users and so on
and so forth. The way we did it for the mastering plugin, it was based on
agreement with existing partners that we had and also human interactions.
- Lander was already based on data that was ethically used, but that was kind of a
closed circle of internal agreements. And now comes the new fair trade program in
the focus. The term is playing with the words of trade mark in the food industry
that is designed to help producers in developing countries achieve sustainable and
equitable trade relationships. But this time, Lander uses it in another context.
We can think of it as a logo put on their AI that tells us it is okay to use
the product as we don't rip off artists. It reminds me a lot of Fairly Trained,
the organization from @NewtonRex, which gives out certificates for tech companies which
use ethically trained AIs. Check out episode 9 of the Iliac Suite in which I talked
about that with Ed. So how has Lander exactly developed that idea?
This company has been built around an idea that we want to democratize access to
music tools and to get people to create more, distribute more and monetize their
work. When we saw the whole AI wave, we said, "Okay, so how do we do it in a
way that's beneficial for the users, for the musicians, and a lot of people around
the team, including our founder Pascal, said how about we do it the most ethically
that we can, and how about we built in what we know best is to give back money
to the musicians. So we design a flow where people can, as we mentioned,
opt in to the program, then they are going to see in a dashboard what revenue
they're getting every month. It can go from a few dollars to a few thousand,
depending on your contribution. And just like you see your earnings from Spotify and
Apple Music, it's in the same type of dashboards and it's super transparent because
you know exactly how much you're making and how it was used. So that notion of
being super transparent, being working for musicians and finding new ways, because
this is a new way to monetize your work, that's how the idea started then of
course our product guys went okay let's try to develop a flow that works let's try
to see how we can make it fit within the existing experience because we didn't want
to disrupt how our users were using our products so we built it into a very very
simple flow you log into your lender account and you can see the different activity
that's that's been on your on your tracks and currently what we're doing is we're
asking our user base and new users to opt in to say, "Hey, I want my track to be
eligible for these upcoming products that you're developing."
These upcoming products are concerning generative AIs, which can create instrumentals
like a baseline or baseline orbit, for example. In three or four months, they are
planning to have enough data to start generating, Frederick told me. But what exactly
do users have to do to be part of that? You have to have a lender account. That's
the minimum. And of course, you have to have distributed a song on our platform,
because this is how you can indicate it's a simple page where it says, make my
songs available or for this fair trade AI program.
It's a tick box, you can opt in, you can opt out. And by the way, I wanna
mention that you're not opted in automatically. You're actually, as a matter of fact,
opted out when you log into your account, but there is a setting or a menu item
there that you can opt in. And then when you say save, it signified to us that
you want to participate. And then in your dashboard, when we're gonna have those
assistive AI tools in the market, you're gonna be able to track and to see the
revenue and in the same type of dashboard that you already have for the distribution
side of things on Spotify and Apple Music and others. So very, very simple process.
You log into your account, you go to that AI setting, you opt in or opt out as
you wish, and then you click save and voila for now. This is where we are today.
If you want to be part of it, of course you have to pay for one of the
subscriptions before. But maybe see it as an investment in many aspects. First of
all, you profit from using all the music making tools Lander has to offer. And now,
secondly, with a new Fairtrade AI program, you can also make money by opting in and
actually not doing anything, maybe sitting on the couch. But how much do you get
paid for that in the end. Angé cannot really predict so far exactly the amount as
they are still in the development. Depending on your contribution, if it's one song,
of course it's going to be less money than if it's a thousand songs and not
automatically are you opted in for all of your songs because some of them might be
just ASMR and stuff like some people use our distribution platform to distribute
sounds and stuff so it has to be eligible and accepted on the lander side. And
then based on your contribution, 20 % of the proceeds of that product that we're
developing, that's not in the market yet, just by the way, just so you know, this
is not mastering plugin. Mastering plugin is already running and it's doing fine. So
it's for upcoming assistive AI tools, 20 % of that. The only thing that we need to
deduct is the server fees like any other tech companies because there are fees
associated with distributing and you know making it available and people can download
it and so on but outside of that 20 % of the proceeds of that product will be put
into your account and you're gonna be able to see it every month and you're gonna
be able to track it over a month depending on your contribution so it's three
things really so it's basically the number of songs that you make available to us
and then it's the 20 % of the proceeds so this is just again minus the server fees
and then ultimately the dashboard that you're gonna be able to track every month.
(upbeat music)
Summing it up, after all expenses, 20 % of the money made with that special
application to generate music will be paid to the participating artists. And from
this percentage, every artist will get money depending on how many songs he or she
has contributed to the dataset. If it comes to opting in and opting out,
so adding music to the dataset, Lander maybe has another problem. What is happening
when people are trying to upload AI generated music to be part of the datasets?
I mean, other companies like Spotify or YouTube know that problem. It cannot be AI
training, AI training, AI. So we can detect that. There's technology involved and
there are also humans. We're musicians here. We know how to listen to music,
how to make it as much as a great product as we can. So there will be checks
within, or as I mentioned, ASMR, as great as ASMR is. Not really useful to train a
product that will be assisting other musicians in building choruses and songs and so
on and so forth. So those are kind of the gate check that we have. - Okay, the
song was tested and approved. How can it be tracked in the AI so that in the end
the user can know for what it was used. Everything is of course tracked throughout
the chain and we can go back to the song, how it was used and what product it
ended up in and what product, what percentage of this product represents within our
ecosystem. So this is all, it's a bit like blockchain I would say, like you can
trace the whole chain back from a song to a product. So very easy to go back then
and say okay so you contributed five songs, ten songs, a thousand songs for this
cohort and it ended up in this product and it represents this percentage within the
Lander ecosystem. Then again very transparently in your account you're going to be
able to see that level of details.
if you think now oh that is interesting I'm a musician I make music I could make
money by sitting at home drinking a coffee while participating and training an AI
with my music so that sounds nice and then you are wondering but is my music
interesting enough for them? Building a song requires vocals, basslines,
drums, guitars and so on and so forth. So basically any type of music is more than
welcome. We love, love, love when there's vocals on the songs because it's one of
the most interesting and one of the most difficult, quite honestly, thing to
reproduce, right? And we saw it with recent, how can We sent evolution on that
front, it's getting very close, very, very good. But it's still something that needs
a lot of training. So all instruments are welcome, all music genre are welcome.
I know there's a lot of hip hop out there, I know there's a lot of rock and
there's a lot of pop and this is what we're already seeing from people opting in.
A lot of those genre are very, very popular right now. Lander is asking your
permission to use your music to train their AI for new generative applications. You
get paid, that is good. You know about it and give your consent, that is very
good. Actually, all that is the opposite of that, what many AI companies out there
do right now, Suno Udio. With that, Lander is taking a big step towards an
ethically used artificial intelligence. On the other side, you must be okay with the
fact that a company copies your style of music and others can then profit from that
style in creating their music. Always keep that in mind. You should feel okay with
that. But I think your music, if it is online streamed, has already been scraped by
tech companies anyway. So why not act, Be active, be proactive by deciding on your
own which company uses your songs and how. Lander also gives you the possibility to
opt out, although that must be done before a new AI starts collecting data because
it's very hard to unlearn or to undo an opt -in within a dataset. So if you did
say yes for the first cohort, of course you are part of it and you'll get
compensated for it. You'll get your revenue associated with it as long as the
product generates proceeds. If you opt out, the next training, you won't be part of
it, the next score, you won't be part of it, and you'll get a confirmation that
you're not part of it. So this is very fair to go back to the name of the
program because you said yes for the first score and you say no for the next ones.
And you can always go out and in. If you say in two cores that, "Oh, you know
what? I want back in. I want to contribute moving forward, you can also opt in
back to the program.
In the end, who are these musicians who should be part of the Fairtrade program of
Lander? Who is it for?
and so forth. Maybe you're like, "Hey, you know what? I want to contribute. I want
to be part of this future trend and this future, and it's not a trend. It's here
to stay. So this future movement, I would say, of music. And my songs are just
sitting there with a couple streams and so on and so forth. And here's a new way
to monetize my work. Not saying that this is going to necessarily make millions of
dollars for musicians, but it's our way to bring as much money back to the
contributors as we weekend. Some last words here before I let Frederik talk one more
time. Of course the company benefits from this because it uses its users data to
further develop its own products and with that they can make profit.
But this happens with the consent and participation of the musicians so both sides
benefit. Although it is not yet exactly clear how much the artist we'll earn it is
a good and logical step, I think. Other AI companies are welcome to adopt that
idea. - At the end of the day, for us, our ethos is we work for musicians and all
of our decisions, whether it's a micro decision on how we're gonna report on AI
earnings versus big decisions like, hey, how do we create an assistive tool that's
powered by AI, but that's not trying to replace a musician? Those are the questions
that companies are asking themselves right now. I cannot talk for other companies.
What I can say is that Lender is highly focused on bringing the industry to a
level where the artist feels that this is the way forward. This is not another,
this is not a tech company trying to monetize the soul all of the artists. It's
quite the opposite. It's, hey, we're there for you. We actually wrote a piece in
Pascal, our founder wrote a piece on Medium, taking the side of the musician,
like because there was a lot of discussion around AI tracks being distributed on
Spotify and so on and so forth. And we're not against AI tracks being distributed
on Spotify, what we are for is getting the musicians compensated for their
contribution. So that is, I think, where the industry is at. There's a lot of
action, very heated debate on how to do it. We think that with fair trade and
other initiatives that we're about to launch in the market is the way to go.
For me, the big word or the big advancement is how do we use it to sparkle more
creativity, more music, bring more people from dreaming about playing music to
streaming, basically on streaming platforms or other touch points. So for me, this is
where the industry is going. And also, there's a big opportunity for the whole
industry to come together and to support artists once and for all and and to think
in every decision they make To think about how do we assist them? Not how are we
replacing them? This is not the goal of at least of our company. That was the new
episode of the iliac suite. Thanks Frederik Rangé from Lander for talking to me. I
hope you enjoyed listening once again to the world of AI and music. You can find
all the other episodes also on my YouTube channel by the same name and of course
on all streaming platforms. Please give me a like if you liked it and of course a
subscription to follow if you want to. If you have any feedback, critics let me
know, you can write me and find my contact in the notes to these episodes, also
the links from the music I used in this episode. Thanks to the ethically trained AI
soundful for that. Thanks for listening. Talk to you again in a month. Take care
and behave.
(upbeat music)