ML Engineer @ CORPUS. PhD in AI & music.
London, UK
Joined March 2022
- Tweets185
- Following192
- Followers536
- Likes3.2K
One possible next step here would be smarter predictions / deeper musical control — tighter steering over voicing, density, and how the model develops ideas across a section.
The full app is there to build on, code is open source.
Fork it and build your own real-time instruments in JUCE :)
Code: github.com/maxgraf96/mrt2-ac…
App & VST3 plugin: github.com/maxgraf96/mrt2-ac…
SA3 Variations: Splice-style sample variations, running entirely on your machine.
Drop in a one-shot or loop and the app will generate a series of musical & timbral variations. Listen to the different outputs and drag the ones you like straight into your DAW.
The same pipeline works for any similar plugin you might want to build.
Code is open source. Fork it, extend it, and build your own apps on top of SA3 in JUCE :)
Next step is breaking open the model for some more fine-grained, musically interesting edits.
Code: github.com/maxgraf96/stable-…
Releases: github.com/maxgraf96/stable-…
App/plugin install guide: github.com/maxgraf96/stable-…
Converted models: huggingface.co/maxgraf/sa3-v…
Demo ↓
Excited to finally share what we’ve been cooking up at CORPUS Music!
Today marks the public release of two apps that I’ve helped shape over the last year:
1. CORPUS Contribution App
This is where musicians can contribute their music to CORPUS.
CMI is also built with privacy in mind: everything runs entirely on our servers, using self-hosted models, with no requests to external AI services.
You can try it here: intelligence.corpus.music/
Huge thanks to everyone in the CORPUS team and our partners who made this possible.
