Skip to content

supersational/JAMMIN-GPT

Repository files navigation

JAMMIN-GPT

Paper | ISMIR Poster

ISMIR Title page

Diagram of operation

Requirements

Optional Requirements:

  • fluidsynth (to use GM soundfonts)
    • run fluidsynth and select it as an output MIDI device in Ableton with 1 channel per track
    • fluidsynth path_to_soundfont/FluidR3_GM.sf2 -pFluidSynth is the command I use

Screenshot

Screenshot of Demo

Add AbletonOSC as a control surface in Ableton

Set as Control Surface

Run the python script

In a seperate terminal window run:

python main.py
  • You should see the output: waiting for named MIDI clip to appear..
  • Now create an empty MIDI clip in Ableton and rename it something like: "piano arp"
    • Create MIDI clip shortcut: cmd+shift+m
    • Rename MIDI clip shortcut: cmd+r
  • If the script works correctly the clip's name should change
  • The MIDI clip should now contain LLM-generated music, make some more!

Debugging

  • Make sure you have added AbletonOSC as a control surface in Ableton
  • Sometimes there is an orange colored message in the bottom status bar of Ableton
  • There are ways to debug remote scripts in Ableton and see print statements, guides can be found online

About

Ableton MIDI-Clip generation using GPT-4

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages