6 ms·

Ask HN: AI-Generated Music?

I'd guess that we would be now ready for great, successful AI-generated music. I know nothing about it, of course - can someone more knowledgeable than me tell me what's going on, what's interesting, what's noteworthy in the year 2022?

kristiandupont 4y ago
There is this: https://www.aiva.ai/ https://www.aiva.ai/ I've toyed around with it a bit. It's impressive for sure, but I am not sure I think of it as anything other than a curiosity.
- bambax 4y ago
  Same. It's very underwhelming for now.
  - p1esk 4y ago
    Yes. I feel like they made zero progress since they launched as far as the actual quality of the music. They added a bunch of product features but not things that matter. Melody and harmony have not improved (still takes ~50 tries to generate something that doesn’t sound mediocre or outright bad.
p1esk 4y ago
https://magenta.tensorflow.org https://magenta.tensorflow.org
kwatsonafter 4y ago
This is a song composed with AI in the style of The Fab. https://www.youtube.com/watch?v=LSHZ_b05W7o https://www.youtube.com/watch?v=LSHZ_b05W7o
elpres 4y ago
There is https://mubert.com https://mubert.com
- elektor 4y ago
  +1 for Mubert. I feel like they've neutered their website but their iOS app is easy to use.
hansword 4y ago
https://stevenwaterman.uk/musetree/ https://stevenwaterman.uk/musetree/ - open front end for MuseNet (by ClosedAI), choose your model characteristics, then right-click on a node to request new bits, choose which you like best, repeat.
pavelegorkin 4y ago
Tinnire https://play.google.com/store/apps/details?id=com.tinnire&hl=en&gl=US https://play.google.com/store/apps/details?id=com.tinnire&hl... Also available on iOS
chris_st 4y ago
Not entirely AI-generated, but one of the two folks on this project used an AI music generator (called "Jukebox", IIRC, but it's not on the page) to generate snippets based on the other's musical ideas, and then the two collaborators worked with those to build songs. Some nice results, here's the "About" page that describes their process: https://ooo.ghostbows.ooo/about/ https://ooo.ghostbows.ooo/about/ ---- Edit: Here's where Robin Sloan mentions Jukebox: https://www.robinsloan.com/lab/notes-on-a-genre/ https://www.robinsloan.com/lab/notes-on-a-genre/
- gwern 4y ago
  Yes, it's Jukebox but it's a strange way to try to get an impression of Jukebox because it's simultaneously heavily edited by them but also leaning heavily into the distortions & artifacts. The sample website would be a much better way to hear Jukebox per se: https://jukebox.openai.com/ https://jukebox.openai.com/
laserdancepony 4y ago
Jean-Michelle Jarre released an app called "EōN" in 2019 that may be fitting to your description. https://apps.apple.com/us/app/e%C5%8Dn-by-jean-michel-jarre/id1477959945 https://apps.apple.com/us/app/e%C5%8Dn-by-jean-michel-jarre/... Article from BBC about the app: https://www.bbc.com/news/entertainment-arts-50335897 https://www.bbc.com/news/entertainment-arts-50335897
riidom 4y ago
https://www.youtube.com/c/DADABOTS_official/about https://www.youtube.com/c/DADABOTS_official/about Death Metal, the only music robots can love.
AriedK 4y ago
https://www.ampermusic.com/ https://www.ampermusic.com/ Amper music, owned by Shutterstock, has been around for a while. It was quite nice to play around with when it was still free, not sure if, and how it has evolved over the past years.
suyash 4y ago
Look up Google Magenta projects. We made a web app using Magenta.JS VAE model : https://devpost.com/software/dear-diary-ezrmgt https://devpost.com/software/dear-diary-ezrmgt app : https://deardiary.ai/ https://deardiary.ai/
spacec0wb0y 4y ago
https://generative.fm/ https://generative.fm/
- bambax 4y ago
  From the home page: > The music from Generative.fm is composed by a human—not AI
sillysaurusx 4y ago
I made a bunch with Gwern’s GPT: https://soundcloud.com/theshawwn/sets/ai-generated-videogame-music https://soundcloud.com/theshawwn/sets/ai-generated-videogame...
- p1esk 4y ago
  Do you think there will be Jukebox2, and similar level of improvement as what we saw with GPT->GPT2->GPT3 progression?
  - sillysaurusx 4y ago
    Undoubtedly. It’s just a matter of time. Whether it’ll be OpenAI or someone else is a tossup though. There’s not much commercial incentive to make AI music. So my money is on the hackers. Intellectual curiosity is still one of the most powerful forces of change. The resources are there. You can get a couple hundred TPUs from google for free. So it’s a matter of talent, not permission.
    - LionTamer 4y ago
      > You can get a couple hundred TPUs from Google for free How?
      - sillysaurusx 4y ago
        TRC (tpu research cloud). Apply and you’ll almost certainly be accepted. Yes, you.
crablegcity899 4y ago
adamnemecek 4y ago
I’m working on something in this space Launching soon https://ngrid.io https://ngrid.io Join the discord https://discord.gg/a5ttYuG https://discord.gg/a5ttYuG
- ilikeitdark 4y ago
  Looks interesting, just signed up and joined the discord. When will you have something to play with?
  - adamnemecek 4y ago
    This month hopefully.
adzm 4y ago
The first step is AI-assisted music, really. And that's a great and useful niche. Think auto accompaniment, backing tracks, etc. Tools for helping figure out transitions. Someone already pointed out magenta which is a great starting point.
rogierhofboer 4y ago
https://www.musicdevelopments.com/ https://www.musicdevelopments.com/ Technically not ‘AI’ though, more based on music theory ‘rules’.
- mdp2021 4y ago
  It is not really clear from those pages how RapidComposer works - but it seems it's still AI (the only condition is that you call melodies etc. "solutions"). Of course there exists rule-based AI. Whatever works to partially replace a human professional.
naillo 4y ago
Found this today, the output is pretty impressive to my ears: https://github.com/lucidrains/perceiver-ar-pytorch https://github.com/lucidrains/perceiver-ar-pytorch + https://storage.googleapis.com/perceiver-ar/index.html https://storage.googleapis.com/perceiver-ar/index.html
- visarga 4y ago
  Great samples.
newaccount74 4y ago
It's an older project, but I love Infinite Gnossienes: https://gnossiennes.mousereeve.com/ https://gnossiennes.mousereeve.com/ It generates an endless version of the famous minimalist piano piece.
bj-rn 4y ago
klingklangklong has a few nice AI/ML based projects, for example: https://klingklangklong.com/project/meandering-river https://klingklangklong.com/project/meandering-river https://klingklangklong.com/project/for-seasons https://klingklangklong.com/project/for-seasons
atomack 4y ago
David Cope's been making algorithmic music for a while - not current but worth knowing about: http://artsites.ucsc.edu/faculty/cope/ http://artsites.ucsc.edu/faculty/cope/ edit: There's also this on ambient endless generative music that I think was an HN submission a couple of years ago https://generative.fm/ https://generative.fm/
balthigor 4y ago
Entirely AI generated music. Lyrics generated by GPT variants, album art by diffusion models, music composed and performed by Jukebox (customized) - http://soundcloud.com/baltigor http://soundcloud.com/baltigor
simple10 4y ago
AIVA has surprisingly good examples on their YouTube playlist that sound like real (human) compositions to my untrained ear. https://www.youtube.com/playlist?list=PLv7BOfa4CxsHAMHQj0ScPXSbgBlLglRPo https://www.youtube.com/playlist?list=PLv7BOfa4CxsHAMHQj0ScP...
winternett 4y ago
One thing that's been pretty interesting is AI vocal splitting, which has revolutionized the art of remixing music that couldn't be fully remixed before. I used it to remix one of my favorite songs ever by Biz Markie just before he passed last year. It ended up being the last remix made of his music before his passing, and I sent it to his manager just before, so I'd like to believe he heard it. It still has a ways to go (as audio quality can be spotty and incomplete), but these online services can often separate more than just vocals now, they can cut individual instruments out of music, and even create pretty good instrumentals. I have been able to remove uncleared vocals from fully mixed tunes that I've made so they can be released as well. I never thought it would have been possible 20 years ago when I started music. As for AI generated music, I think it will be a travesty to de-value or remove humans from the music/art making process entirely, it will always likely be something derivative of human work in essence anyway, but I don't think it will ever match the depth and soul of human-generated music to people who truly know and love music, some things just can't be emulated. Here's the remix video - https://www.youtube.com/watch?v=xL14JH5f-qM https://www.youtube.com/watch?v=xL14JH5f-qM
- kyriakos 4y ago
  which service/software did you use for splitting?
Rochus 4y ago
Here is a recent, interesting survey paper: https://hal.archives-ouvertes.fr/hal-01660772 https://hal.archives-ouvertes.fr/hal-01660772 From my perspective as a trained, practicing musician MuseNet produces quite credible results: https://openai.com/blog/musenet/ https://openai.com/blog/musenet/. But we're not yet there; AI generated music is still recognizable as such. Another, earlier project with very good results was https://www.microsoft.com/en-us/research/publication/automatic-stylistic-composition-of-bach-chorales-with-deep-lstm/ https://www.microsoft.com/en-us/research/publication/automat...; unfortunately the website is no longer active.
nobodywasishere 4y ago
Here's 10 hours of prorcedurally generated djent https://m.youtube.com/watch?v=52SRWJ2pMXI https://m.youtube.com/watch?v=52SRWJ2pMXI
a-dub 4y ago
a few years ago i went to a talk by some people that were working on this kind of stuff within google brain. unfortunately i don't remember any names though. edit: just had a look. it appears to now be the magenta project which is referenced elsewhere in this thread.
songeater 4y ago
I've been working with OpenAi's jukebox to make vocals / "singer-songwriter" type stuff rather than electronica. Can do a HN writeup on process if there is interest... Youtube Playlist: https://www.youtube.com/watch?v=_lcCJzfXl50&list=PLaa32nLgVvrkyveSU8rtD4l02eYQtRlWA https://www.youtube.com/watch?v=_lcCJzfXl50&list=PLaa32nLgVv... SoundCloud: https://soundcloud.com/songshtr/sets/machinecroon https://soundcloud.com/songshtr/sets/machinecroon Github: https://songshtr.github.io https://songshtr.github.io The main challenge for me so far has been post-processing. Jukebox produces very "scratchy" / lofi outputs... and I've been DIY'ing with Audacity. Thinking of getting a track professionally mastered to see just how far the boundary is...
- YetAnotherNick 4y ago
  That's impressive. And the music is much more different than I expected, in a good way. So you wrote the lyrics? Or is it autogenerated as well?
  - songeater 4y ago
    Thanks so much for your kind words... and for taking the time to listen! Yes I wrote the lyrics myself (other than a couple quotes from Eliot & Borges) - no GPT etc involved in the text. Even with the tunes, the seed music fed to Jukebox gives some control over what gets generated next... especially when you later splice various outputs together. Each ~2-3 minute song needs anywhere between 15-100 2-3 minute raw snippets to be generated from Jukebox before you can find enough sections to paste together. I suppose I would be considered a writer/producer while Jukebox is clearly the performer - an immensely talented, yet completely capricious one!
    - 8bitsrule 4y ago
      Neat! Wondering where the singing voices came from, thought I heard some Cash- and Dylan- like phrasing, dialect in some places.
- simonebrunozzi 4y ago
  Yes, would love a writeup!
impetus1 4y ago
Funnily enough, there is something on quitrit(quantum computing) generated sound representation which does generate music. https://github.com/scottoshiro2/IBM_QCHack_Challenge2021 https://github.com/scottoshiro2/IBM_QCHack_Challenge2021
shannifin 4y ago
I'm actively working on a startup focusing on this (while doing startup school this month). For now I'm focusing on the melody generation, as to me that seems to be the area most lacking in AI-generated music, yet what I find the most interesting to work with and explore as a composer. That said, I don't have anything to show yet... Only got a landing page at https://tunesage.com https://tunesage.com at the moment... but here's an ancient prototype writing very generic melodies just to prove I'm trying... https://www.youtube.com/watch?v=8cwEaiwEZfU https://www.youtube.com/watch?v=8cwEaiwEZfU Regardless of my personal eventual success or failure in the endeavour, the field does seem to be ripe for innovation! Nothing out there at the moment (that I've seen at least) seems to satisfy my current desires as a composer.
smugma 4y ago
UK government seems to be supporting it (AI qualifying for copyright): https://news.ycombinator.com/item?id=31920524 https://news.ycombinator.com/item?id=31920524
WithinReason 4y ago
Here is one of my favourite YouTube channels, Two Set Violin (professional musicians) listening to an AI generated symphony: https://www.youtube.com/watch?v=R69JYEfCSeI https://www.youtube.com/watch?v=R69JYEfCSeI
peterlk 4y ago
Has anyone solved the data problem? I have taken many stabs at this over the last 10 years, and the primary problem is that music is _extremely_ subtle. Even something as seemingly simple as chord progressions are very difficult to annotate properly without deep training. And even then, experts can disagree. The thing that's missing is the data. If we had midi transcriptions of 100k songs (abc notation could be fine too), we could probably get really interesting stuff, but most of what is available is lossy chord transcriptions and classical music (public domain). So if you want to automatically create something that sounds like mozart, you're in luck! But this isn't really satisfying to me. For generative music, we're still largely stuck with encoding musical rules in code rather than feeding data to a transformer. To me, the former feels much less like AI than the latter. The data is all locked up behind an impossible quagmire of copyright. But if I were a sheet music publishing company, I would be seriously considering the future of music creation with AI given my broad access to notated music & metadata (is this an original score, or a grade 1 simplification?). But again, music copyright is a pretty complex contraption.
- songeater 4y ago
  >> has anyone solved the data problem? So MIDI songs were being generated quite "early" on, even with basic text generators (LSTM models), where the "alphabet" was replaced by MIDI symbols. [1] What Jukebox (and other models) did well was work with raw audio, rather than rely on MIDI or similar transcriptions. They break down audio into "blocks" using Fast Fourier Transforms, then each block is "tokenized" and then fed into transformers, similar to GPT-3 or other newer text generation models. This now allows the "subtle" musicalities to be discovered by models without relying on transcription. Wavenet (from Google) [2] I believe was the first one to do this... I tried myself for 6 months, but the processing power required for this is candidly only available for the Google/OpenAIs of the world. >> the data is all locked up behind an impossible quagmire of copyright So my understanding is that the training of Jukebox is similar to Copilot. It was trained on the entire English Spotify catalogue so all that has been digested by the model - under Fair Use. But again - this is probably far from "settled" legally. [1] https://arxiv.org/abs/1612.07837 https://arxiv.org/abs/1612.07837 [2] https://www.deepmind.com/blog/wavenet-a-generative-model-for-raw-audio https://www.deepmind.com/blog/wavenet-a-generative-model-for...
Ashbt 4y ago
Made a bunch a while ago using Jukebox and cleaned it up a bit using some izotope plugins: https://m.soundcloud.com/R05c0 https://m.soundcloud.com/R05c0 It's really just a matter of time before they figure out how to make the music a bit more "cohesive." Right now it's kinda like a stream of consciousness, vs. being verse, chorus, verse, chorus. My prediction: in under 5 years, most of us will have a favorite AI band.
alxmrs 4y ago
I would love to see a LM generate sonic-pi scripts which generates music.
charneykaye 4y ago
Hiya! I'm the founder of XJ music. It's taken me seven years to build this thing I dreamed of when I was young, producing electronic music on a pair of MPC-2000s trying to hack them into something more. XJ is human created content fed into a "player" that fuses the composition process with the storage and distribution. It's an algorithmic medium. The music is never "done" and what you hear in the app is playing live in real time from the software we run in the cloud. What do you think?? Free on iOS and Android. https://xj.io/download https://xj.io/download
emperorcj 4y ago
https://dadabots.com/ https://dadabots.com/ https://www.youtube.com/watch?v=MwtVkPKx3RA https://www.youtube.com/watch?v=MwtVkPKx3RA

kristiandupont 4y ago

There is this: https://www.aiva.ai/ https://www.aiva.ai/ I've toyed around with it a bit. It's impressive for sure, but I am not sure I think of it as anything other than a curiosity.

bambax 4y ago
Same. It's very underwhelming for now.
- p1esk 4y ago
  Yes. I feel like they made zero progress since they launched as far as the actual quality of the music. They added a bunch of product features but not things that matter. Melody and harmony have not improved (still takes ~50 tries to generate something that doesn’t sound mediocre or outright bad.

p1esk 4y ago

https://magenta.tensorflow.org https://magenta.tensorflow.org

kwatsonafter 4y ago

This is a song composed with AI in the style of The Fab. https://www.youtube.com/watch?v=LSHZ_b05W7o https://www.youtube.com/watch?v=LSHZ_b05W7o

elpres 4y ago

There is https://mubert.com https://mubert.com

elektor 4y ago
+1 for Mubert. I feel like they've neutered their website but their iOS app is easy to use.

hansword 4y ago

https://stevenwaterman.uk/musetree/ https://stevenwaterman.uk/musetree/ - open front end for MuseNet (by ClosedAI), choose your model characteristics, then right-click on a node to request new bits, choose which you like best, repeat.

pavelegorkin 4y ago

Tinnire https://play.google.com/store/apps/details?id=com.tinnire&hl=en&gl=US https://play.google.com/store/apps/details?id=com.tinnire&hl... Also available on iOS

chris_st 4y ago

Not entirely AI-generated, but one of the two folks on this project used an AI music generator (called "Jukebox", IIRC, but it's not on the page) to generate snippets based on the other's musical ideas, and then the two collaborators worked with those to build songs. Some nice results, here's the "About" page that describes their process: https://ooo.ghostbows.ooo/about/ https://ooo.ghostbows.ooo/about/ ---- Edit: Here's where Robin Sloan mentions Jukebox: https://www.robinsloan.com/lab/notes-on-a-genre/ https://www.robinsloan.com/lab/notes-on-a-genre/

gwern 4y ago
Yes, it's Jukebox but it's a strange way to try to get an impression of Jukebox because it's simultaneously heavily edited by them but also leaning heavily into the distortions & artifacts. The sample website would be a much better way to hear Jukebox per se: https://jukebox.openai.com/ https://jukebox.openai.com/

laserdancepony 4y ago

Jean-Michelle Jarre released an app called "EōN" in 2019 that may be fitting to your description. https://apps.apple.com/us/app/e%C5%8Dn-by-jean-michel-jarre/id1477959945 https://apps.apple.com/us/app/e%C5%8Dn-by-jean-michel-jarre/... Article from BBC about the app: https://www.bbc.com/news/entertainment-arts-50335897 https://www.bbc.com/news/entertainment-arts-50335897

riidom 4y ago

https://www.youtube.com/c/DADABOTS_official/about https://www.youtube.com/c/DADABOTS_official/about Death Metal, the only music robots can love.

AriedK 4y ago

https://www.ampermusic.com/ https://www.ampermusic.com/ Amper music, owned by Shutterstock, has been around for a while. It was quite nice to play around with when it was still free, not sure if, and how it has evolved over the past years.

suyash 4y ago

Look up Google Magenta projects. We made a web app using Magenta.JS VAE model : https://devpost.com/software/dear-diary-ezrmgt https://devpost.com/software/dear-diary-ezrmgt app : https://deardiary.ai/ https://deardiary.ai/

spacec0wb0y 4y ago

https://generative.fm/ https://generative.fm/

bambax 4y ago
From the home page: > The music from Generative.fm is composed by a human—not AI

sillysaurusx 4y ago

I made a bunch with Gwern’s GPT: https://soundcloud.com/theshawwn/sets/ai-generated-videogame-music https://soundcloud.com/theshawwn/sets/ai-generated-videogame...

p1esk 4y ago
Do you think there will be Jukebox2, and similar level of improvement as what we saw with GPT->GPT2->GPT3 progression?
- sillysaurusx 4y ago
  Undoubtedly. It’s just a matter of time. Whether it’ll be OpenAI or someone else is a tossup though. There’s not much commercial incentive to make AI music. So my money is on the hackers. Intellectual curiosity is still one of the most powerful forces of change. The resources are there. You can get a couple hundred TPUs from google for free. So it’s a matter of talent, not permission.
  - LionTamer 4y ago
    > You can get a couple hundred TPUs from Google for free How?
    - sillysaurusx 4y ago
      TRC (tpu research cloud). Apply and you’ll almost certainly be accepted. Yes, you.

crablegcity899 4y ago

adamnemecek 4y ago

I’m working on something in this space Launching soon https://ngrid.io https://ngrid.io Join the discord https://discord.gg/a5ttYuG https://discord.gg/a5ttYuG

ilikeitdark 4y ago
Looks interesting, just signed up and joined the discord. When will you have something to play with?
- adamnemecek 4y ago
  This month hopefully.

adzm 4y ago

The first step is AI-assisted music, really. And that's a great and useful niche. Think auto accompaniment, backing tracks, etc. Tools for helping figure out transitions. Someone already pointed out magenta which is a great starting point.

rogierhofboer 4y ago

https://www.musicdevelopments.com/ https://www.musicdevelopments.com/ Technically not ‘AI’ though, more based on music theory ‘rules’.

mdp2021 4y ago
It is not really clear from those pages how RapidComposer works - but it seems it's still AI (the only condition is that you call melodies etc. "solutions"). Of course there exists rule-based AI. Whatever works to partially replace a human professional.

naillo 4y ago

Found this today, the output is pretty impressive to my ears: https://github.com/lucidrains/perceiver-ar-pytorch https://github.com/lucidrains/perceiver-ar-pytorch + https://storage.googleapis.com/perceiver-ar/index.html https://storage.googleapis.com/perceiver-ar/index.html

visarga 4y ago
Great samples.

newaccount74 4y ago

It's an older project, but I love Infinite Gnossienes: https://gnossiennes.mousereeve.com/ https://gnossiennes.mousereeve.com/ It generates an endless version of the famous minimalist piano piece.

bj-rn 4y ago

klingklangklong has a few nice AI/ML based projects, for example: https://klingklangklong.com/project/meandering-river https://klingklangklong.com/project/meandering-river https://klingklangklong.com/project/for-seasons https://klingklangklong.com/project/for-seasons

atomack 4y ago

David Cope's been making algorithmic music for a while - not current but worth knowing about: http://artsites.ucsc.edu/faculty/cope/ http://artsites.ucsc.edu/faculty/cope/ edit: There's also this on ambient endless generative music that I think was an HN submission a couple of years ago https://generative.fm/ https://generative.fm/

balthigor 4y ago

Entirely AI generated music. Lyrics generated by GPT variants, album art by diffusion models, music composed and performed by Jukebox (customized) - http://soundcloud.com/baltigor http://soundcloud.com/baltigor

simple10 4y ago

AIVA has surprisingly good examples on their YouTube playlist that sound like real (human) compositions to my untrained ear. https://www.youtube.com/playlist?list=PLv7BOfa4CxsHAMHQj0ScPXSbgBlLglRPo https://www.youtube.com/playlist?list=PLv7BOfa4CxsHAMHQj0ScP...

winternett 4y ago

One thing that's been pretty interesting is AI vocal splitting, which has revolutionized the art of remixing music that couldn't be fully remixed before. I used it to remix one of my favorite songs ever by Biz Markie just before he passed last year. It ended up being the last remix made of his music before his passing, and I sent it to his manager just before, so I'd like to believe he heard it. It still has a ways to go (as audio quality can be spotty and incomplete), but these online services can often separate more than just vocals now, they can cut individual instruments out of music, and even create pretty good instrumentals. I have been able to remove uncleared vocals from fully mixed tunes that I've made so they can be released as well. I never thought it would have been possible 20 years ago when I started music. As for AI generated music, I think it will be a travesty to de-value or remove humans from the music/art making process entirely, it will always likely be something derivative of human work in essence anyway, but I don't think it will ever match the depth and soul of human-generated music to people who truly know and love music, some things just can't be emulated. Here's the remix video - https://www.youtube.com/watch?v=xL14JH5f-qM https://www.youtube.com/watch?v=xL14JH5f-qM

kyriakos 4y ago
which service/software did you use for splitting?

Rochus 4y ago

Here is a recent, interesting survey paper: https://hal.archives-ouvertes.fr/hal-01660772 https://hal.archives-ouvertes.fr/hal-01660772 From my perspective as a trained, practicing musician MuseNet produces quite credible results: https://openai.com/blog/musenet/ https://openai.com/blog/musenet/. But we're not yet there; AI generated music is still recognizable as such. Another, earlier project with very good results was https://www.microsoft.com/en-us/research/publication/automatic-stylistic-composition-of-bach-chorales-with-deep-lstm/ https://www.microsoft.com/en-us/research/publication/automat...; unfortunately the website is no longer active.

nobodywasishere 4y ago

Here's 10 hours of prorcedurally generated djent https://m.youtube.com/watch?v=52SRWJ2pMXI https://m.youtube.com/watch?v=52SRWJ2pMXI

a-dub 4y ago

a few years ago i went to a talk by some people that were working on this kind of stuff within google brain. unfortunately i don't remember any names though. edit: just had a look. it appears to now be the magenta project which is referenced elsewhere in this thread.

songeater 4y ago

I've been working with OpenAi's jukebox to make vocals / "singer-songwriter" type stuff rather than electronica. Can do a HN writeup on process if there is interest... Youtube Playlist: https://www.youtube.com/watch?v=_lcCJzfXl50&list=PLaa32nLgVvrkyveSU8rtD4l02eYQtRlWA https://www.youtube.com/watch?v=_lcCJzfXl50&list=PLaa32nLgVv... SoundCloud: https://soundcloud.com/songshtr/sets/machinecroon https://soundcloud.com/songshtr/sets/machinecroon Github: https://songshtr.github.io https://songshtr.github.io The main challenge for me so far has been post-processing. Jukebox produces very "scratchy" / lofi outputs... and I've been DIY'ing with Audacity. Thinking of getting a track professionally mastered to see just how far the boundary is...

YetAnotherNick 4y ago
That's impressive. And the music is much more different than I expected, in a good way. So you wrote the lyrics? Or is it autogenerated as well?
- songeater 4y ago
  Thanks so much for your kind words... and for taking the time to listen! Yes I wrote the lyrics myself (other than a couple quotes from Eliot & Borges) - no GPT etc involved in the text. Even with the tunes, the seed music fed to Jukebox gives some control over what gets generated next... especially when you later splice various outputs together. Each ~2-3 minute song needs anywhere between 15-100 2-3 minute raw snippets to be generated from Jukebox before you can find enough sections to paste together. I suppose I would be considered a writer/producer while Jukebox is clearly the performer - an immensely talented, yet completely capricious one!
  - 8bitsrule 4y ago
    Neat! Wondering where the singing voices came from, thought I heard some Cash- and Dylan- like phrasing, dialect in some places.
simonebrunozzi 4y ago
Yes, would love a writeup!

impetus1 4y ago

Funnily enough, there is something on quitrit(quantum computing) generated sound representation which does generate music. https://github.com/scottoshiro2/IBM_QCHack_Challenge2021 https://github.com/scottoshiro2/IBM_QCHack_Challenge2021

shannifin 4y ago

I'm actively working on a startup focusing on this (while doing startup school this month). For now I'm focusing on the melody generation, as to me that seems to be the area most lacking in AI-generated music, yet what I find the most interesting to work with and explore as a composer. That said, I don't have anything to show yet... Only got a landing page at https://tunesage.com https://tunesage.com at the moment... but here's an ancient prototype writing very generic melodies just to prove I'm trying... https://www.youtube.com/watch?v=8cwEaiwEZfU https://www.youtube.com/watch?v=8cwEaiwEZfU Regardless of my personal eventual success or failure in the endeavour, the field does seem to be ripe for innovation! Nothing out there at the moment (that I've seen at least) seems to satisfy my current desires as a composer.

smugma 4y ago

UK government seems to be supporting it (AI qualifying for copyright): https://news.ycombinator.com/item?id=31920524 https://news.ycombinator.com/item?id=31920524

WithinReason 4y ago

Here is one of my favourite YouTube channels, Two Set Violin (professional musicians) listening to an AI generated symphony: https://www.youtube.com/watch?v=R69JYEfCSeI https://www.youtube.com/watch?v=R69JYEfCSeI

peterlk 4y ago

Has anyone solved the data problem? I have taken many stabs at this over the last 10 years, and the primary problem is that music is _extremely_ subtle. Even something as seemingly simple as chord progressions are very difficult to annotate properly without deep training. And even then, experts can disagree. The thing that's missing is the data. If we had midi transcriptions of 100k songs (abc notation could be fine too), we could probably get really interesting stuff, but most of what is available is lossy chord transcriptions and classical music (public domain). So if you want to automatically create something that sounds like mozart, you're in luck! But this isn't really satisfying to me. For generative music, we're still largely stuck with encoding musical rules in code rather than feeding data to a transformer. To me, the former feels much less like AI than the latter. The data is all locked up behind an impossible quagmire of copyright. But if I were a sheet music publishing company, I would be seriously considering the future of music creation with AI given my broad access to notated music & metadata (is this an original score, or a grade 1 simplification?). But again, music copyright is a pretty complex contraption.

songeater 4y ago
>> has anyone solved the data problem? So MIDI songs were being generated quite "early" on, even with basic text generators (LSTM models), where the "alphabet" was replaced by MIDI symbols. [1] What Jukebox (and other models) did well was work with raw audio, rather than rely on MIDI or similar transcriptions. They break down audio into "blocks" using Fast Fourier Transforms, then each block is "tokenized" and then fed into transformers, similar to GPT-3 or other newer text generation models. This now allows the "subtle" musicalities to be discovered by models without relying on transcription. Wavenet (from Google) [2] I believe was the first one to do this... I tried myself for 6 months, but the processing power required for this is candidly only available for the Google/OpenAIs of the world. >> the data is all locked up behind an impossible quagmire of copyright So my understanding is that the training of Jukebox is similar to Copilot. It was trained on the entire English Spotify catalogue so all that has been digested by the model - under Fair Use. But again - this is probably far from "settled" legally. [1] https://arxiv.org/abs/1612.07837 https://arxiv.org/abs/1612.07837 [2] https://www.deepmind.com/blog/wavenet-a-generative-model-for-raw-audio https://www.deepmind.com/blog/wavenet-a-generative-model-for...

Ashbt 4y ago

Made a bunch a while ago using Jukebox and cleaned it up a bit using some izotope plugins: https://m.soundcloud.com/R05c0 https://m.soundcloud.com/R05c0 It's really just a matter of time before they figure out how to make the music a bit more "cohesive." Right now it's kinda like a stream of consciousness, vs. being verse, chorus, verse, chorus. My prediction: in under 5 years, most of us will have a favorite AI band.

alxmrs 4y ago

I would love to see a LM generate sonic-pi scripts which generates music.

charneykaye 4y ago

Hiya! I'm the founder of XJ music. It's taken me seven years to build this thing I dreamed of when I was young, producing electronic music on a pair of MPC-2000s trying to hack them into something more. XJ is human created content fed into a "player" that fuses the composition process with the storage and distribution. It's an algorithmic medium. The music is never "done" and what you hear in the app is playing live in real time from the software we run in the cloud. What do you think?? Free on iOS and Android. https://xj.io/download https://xj.io/download

emperorcj 4y ago

https://dadabots.com/ https://dadabots.com/ https://www.youtube.com/watch?v=MwtVkPKx3RA https://www.youtube.com/watch?v=MwtVkPKx3RA