open source ai music tools for musicians.
gary used to be a max4live device with 3 models
VST3 / AU / standalone
eight open source ai music models and a native stem separator, in a single vst. put it on your master, press play, record some audio, and start iterating.
it's for musicians who want ai inside the session, working on the track they already have. if you just want pure text-to-music, you probably don't need a vst.
made in the plugin
yuey and carey, then sa3
yuey with ACE-Step and a billie LoRA. sa3 takes over at 0:47.
four sa3 LoRAs
one session, four LoRAs. the header shows which one is driving.
for the phone
make music on your phone with sa3, musicgen, and melodyflow.
v4 is live on android and ios. we’re still iterating.
sa3 · 013
gary4beatbox v4
also from the patch
run models and train your own LoRAs locally
text to playable synth. runs on sa3.cpp.
stable audio 3 fully embedded inside ableton’s context menu. no external apps. all c++. runs on sa3.cpp.
stable audio 3 in VST3, AU, REAPER extension + standalone. no companion app. all c++. runs on sa3.cpp. bring your own LoRAs.
this site started out as a radio station
all of it made with gary. everyone who tunes in hears the same thing, and it keeps playing while you look around.
