- Generate Similar Audio : Jukebox
- Continue with Similar Audio : AudioLM
- Generate Similar Midi : Magenta (custom-training exple), MuseNet
- Transfer Style (Midi) : MuseNet ?
- Transfer Rythmic Style : Groove2Groove
- Interpolate 2 Sources : Magenta Studio Interpolate
- Transfer/Morph Audio tone/timber : DDSP, IRCAM Acids tools ?
- Clone to playable Midi instrument : midi-ddsp, midi2params, Mawf
- Organize/Explore Large Audio Samples (local) collection : Sononym, Algonaut.audio
- Separate Audio into its (sub-Audio) Instruments (= Unmixing) : demucs (better), spleeter (faster)
- Decompose Audio (Spectral/Melodic) : Factorsynth, DeepCreate
- Convert Audio (Complex/multi-track) to Midi (= Transcription) : magenta/mt3, omnizart
- Convert Audio Instruments to Midi (more precise, piano here) : Onsets and Frames, bytedance/piano_transcription
These are mostly still experimental/early stage uses of AI. I only included tools that augment creative music creation. So, I did not consider neither generic Music generation (one click composition/sountrack) nor Music (post-)production (mixing/mastering).
You can find some of these in this spreadsheet (not mine) .
For the[ AI + Music Biz ] kind of news, you can follow AI-Music / AI / Artificial Intelligence tags on musically.