[HN Gopher] Open music foundation models for full-song generation
___________________________________________________________________
Open music foundation models for full-song generation
Author : selvan
Score : 53 points
Date : 2025-08-04 10:46 UTC (3 days ago)
(HTM) web link (map-yue.github.io)
(TXT) w3m dump (map-yue.github.io)
| lotyrin wrote:
| Very nice. Anyone know of projects that aren't tackling the full-
| song problem but rather instrument parts/loops/stems/acapellas?
| I'd like something that's more like "infinite AI
| Loopcloud/Splice" most of these full-song models don't do well to
| be asked for individual parts in my experience (though I will
| have to try it with this one).
| platers wrote:
| https://suno.com/studio-waitlist Just a waitlist so far, but
| looks like this is the direction suno is going
| lotyrin wrote:
| Yeah... I hope this is what their plan is with that, but I'm
| not entirely certain.
| rwmj wrote:
| Also live AI dueting would be interesting, like having a
| virtual guitarist you could jam/duet with.
| lotyrin wrote:
| Yeah. Or like, a loop that plays continuously and has style
| parameters exposed you can tweak with a controller like a
| Midi Fighter Twister and get feedback from in real-time. Then
| you could do something akin to DJ/live production by having
| two of these going in sync with each other into a mixer.
| (Tweak params of the cue track until you like it, transition
| at a phrase point, repeat).
| HxokcPwi wrote:
| Like this? https://aistudio.google.com/apps/bundled/promptd
| j?showPrevie...
| HxokcPwi wrote:
| Just saw this today:
| https://x.com/jesseengel/status/1953496623696556478
| HxokcPwi wrote:
| Try https://magenta.withgoogle.com/infinite-crate
| vunderba wrote:
| This gets discussed a lot but unfortunately there's just not
| much out there around this.
|
| The closest thing I've seen is virtual drummers in Logic X
| which will follow along with the structure of your song and
| generate a percussive accompaniment. It's no substitute for a
| real drummer but it's serviceable.
| ssalka wrote:
| Something interesting... the first 10 seconds or so of the "Death
| Growl" example[1] is basically copied _verbatim_ from "Ov Fire
| And The Void" by Behemoth.
|
| More specifically, I think the part that seems copied is at 2:13
| of the original[2], as it leads into a solo-ish bit which in the
| AI version sounds similar still, but goes on to do its own thing:
|
| [1] https://map-yue.github.io/music/moon.death_metal.mp3
|
| [2] https://youtu.be/vAmnsKKrt9w?t=133
| someothherguyy wrote:
| > Additionally, our memorization-effect experiments in Section
| 11 demonstrate that our design maintains creativity without
| plagiarizing, even under strong training set conditioning.
|
| https://arxiv.org/html/2503.08638v1#S11
| amelius wrote:
| Does Shazam think it is the same?
| vorgol wrote:
| The youtube link is suddenly not available any more (at least
| in the UK)
| bangaladore wrote:
| What is the use case for music generation models? I see usecases
| for alot of the other foundation models like text, image, tts,
| sst, but why do I want AI generated music?
| FridgeSeal wrote:
| Now you don't need to know how to make music! You're finally
| free of all those pesky, elitist musicians gate-keeping
| music!!!!1!
| frank_nitti wrote:
| I've mostly used them for laughs with my friends. Sometimes
| generating "custom" songs with funny lyrics, but most fun so
| far is editing lyrics of existing songs to say ridiculous
| things for fun.
|
| No real clue how someone would use them for a more serious
| endeavor, only thing I could imagine would be to quickly
| iterate/prototype with song structures on a fixed seed to
| generate ideas for a real composition. Consider the case of an
| indie game developer or film maker getting some placeholder
| music to test the experience during early throwaway iterations.
| libraryatnight wrote:
| Generating crappy background music for reality TV?
| scarecrowbob wrote:
| yeah, but have yall made any progress in a model that can have
| sex with my partner for me?
___________________________________________________________________
(page generated 2025-08-07 23:00 UTC)