fp.

The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device.

The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.

Comments

yoouareperfect•1h ago

This is really awesome thanks for sharing

evalystai•1h ago

The idea is awesome! :) However there's definitely much room for improvement, first of all rythm and composition (so there's some sense of musical form).

simedw•1h ago

Thank you.

Yes, I think I’ve gotten it to roughly a GPT-2 level: good enough to share, but with a lot of room left to improve. I think adding some kind of bar/measure token might help with rhythm, and perhaps some form of longer-term planning for the overall composition.

isoprophlex•1h ago

Amazing idea! Gonna hook this up to my little synthesizer and blast some square wave arpeggiated ML music!

heikkilevanto•32m ago

I would love something like that, except that I play the melody, and it produces proper 3-4 part accompaniment, preferably in good baroque style. Extra bonus if it could also write it into a file in a format suitable for music editing programs.

morkalork•25m ago

That's a fun idea. You could start playing the piano and it kicks in with a base and drums for a jazz band.

subhajeet2107•26m ago

This is so amazing, can you improve the quality of generation at the cost of notes per seconds ? No one can play 108 notes/sec anyways, maybe you can train the model to do CoT for better quality

devonsolomon•22m ago

This is really fun. Scaler 3 starts with a chord progression and lets you break it down into musical performances and parts. Useful for ideation when producing.

Would be fun to get a midi clock going and play some chords on my piano and have my synth start jamming along with the bass and my keyboard doing some performance. Or any combination of the above.

mh-•14m ago

+1 all of this. That would be incredible (this already seems very cool - excited to get home and try it!)

pil0u•20m ago

Even after a few years deep into AI, I find your application absolutely magic. This is very inspiring, thank you for sharing.

timmb•17m ago

Reminds me of Francois Pachet’s Continuator (all the way back in 2003, using hierarchical markov models)

https://www.francoispachet.fr/continuator/

leobg•16m ago

I don’t have MIDI. How about whistling or playing the piano via microphone? Sounds easy. Another 6 month rabbit hole? :)

goda90•14m ago

Reminds me of this project to generate every melody possible algorithmically in order to fight music copyright lawsuits. https://allthemusic.info/

jwr•13m ago

> Eventually I used Gemini 3.5 Flash for pairwise evaluation

But, but… wouldn't that be… (gasp) DISTILLATION?

Fun project!

ramesh31•13m ago

But does it work, or does it just output mush after a few seconds? Current SOTA in this domain is https://github.com/EleutherAI/aria a Llama 3.2 based 1B model trained on https://github.com/bytedance/GiantMIDI-Piano which will trivially run on an iPhone these days. It's results are... pretty much useless too.

davidajackson•10m ago

How would you expand this to support elements like attack ("velocity of the key-down" in piano speak), grace notes, timing etc. Would each of those be part of this model or another model? How would you model an arbitrary element (pedal, duration, etc...)