YuE2 Generates Full Songs with Symbolic Planning
YuE2 · Frontier Music with Symbolic Planning
YuE2 is a music generation model that creates complete songs with symbolic planning. It generates both audio and symbolic scores (ABC notation) for each piece, allowing users to explore melody, rhythm, and chords. The demo showcases 65 cases across styles like heavy metal, jazz-funk, and chamber folk, with transformations such as tempo changes, lyric rewrites, and style rearrangements.
Listen to a song, then explore the melody, rhythm, and chords in its symbolic plan.
- hypersoar
I mentioned this in another comment thread, recently, but I value art, including music, for the fact that another human being(s) created it to express something. I have zero interest in it if it's AI-generated. The thought of endless content created from the void for me to gorge on like oreos fills me with nothing but despair.
- stevage
It's really depressing that I have gotten into music composition right when this is happening.
All of these projects are basically just destroying entire fields of human creativity.
It doesn't seem so bad at first when the quality is mediocre, but as soon as they get good, then any human composition will be completely devalued, despite the enormous amount of knowledge, skill and effort that goes into producing one.
A composer who writes a piece of music themselves, instead of being recognised and admired basically becomes the equivalent of someone who bakes their own sourdough instead of going to the supermarket: a curiosity, someone who does things the hard way for their own amusement.
Graphic artists and bloggers were the first to suffer this, but it feels like it's just going to happen to basically every type of creative expression that can be represented digitally. It's just a question of how many months until that happens.
- vunderba
This seems to be one of the few open-weight models that isn't just another "text to audio" generator. You can take existing sheet music, convert it to ABC notation, and then experiment with various covers.
I don't have a ton of time right now to really mess around with it, but I took 15 minutes to run it on my Mac M1 with an old melody I wrote back in high school for an RPG I was writing in VB6 (dating myself a bit here).
If anyone wants to hear what it sounds like:
- tom_vidal
We are creating animated corpses out of our culture.
The point of art isn’t to generate beautiful patterns. It’s to create meaning. And you can’t have that without human care and intent.
Furthermore, that care needs to be intrinsically interwoven into the patterns that comprise the work. That’s what craft is. The mastery to generate exquisite patterns, and the care to infuse them with meaning at every level.
Without care, in a Heideggerian sense, we are left with something which tastes sweet but is completely hollow, perhaps even illusional.
- dcsommer
For an informed take (from a professional musician, composer, and arranger) on where this is going for musicians and the music industry, I really recommend Adam Neely's video essay "Suno, AI Music, and the Bad Future": https://www.youtube.com/watch?v=U8dcFhF0Dlk
- dr_kiszonka
The vocals in the few songs I listened to are so very clearly AI. I find Suno's much more credible in many genres.
If anyone has tried Suno's Studio 2, I would love to hear your take on it. At $24-30/month, it's approaching Ableton Live 12 Suite rent-to-own territory, without the "own" part.
- euleriancon
I've been waiting for symbolic music generators for ages. This is so nice for musicians. Being able to intelligently translate parts, generate additional parts, generate sheet music from audio, generate variations on pieces, making pieces easier or harder.
So many possibilities.
- mg1c
Taste is still what separates AI output from human output regardless of whether it's music or software, IMO. Thinking in modes vs. functional harmony, how you hear a chord in context, which voicing you reach for etc. are things that can hardly be abstracted into strict music theory, let alone pattern-matched.
I imagine you can get pretty close to replicating an artist's sound if you specifically train on their output, but at that point it's the same question of whether you're satisfied with abstracting away the things that make writing music/software interesting to us (with music the answer's a lot more obvious since the point of writing music has always just been... to write music).