A software solution that facilitates a collaborative musical experience between human musicians and generative AI.
- I want to be able to jam with a generative AI model in much the same way that you would practice with other humans.
- I explain to the system what kind of mood I'm in or what style of music I plan to experiment with in natural language. It then uses a pre-trained model to kick off some session drums or whatever accompaniment I prefer or that seems appropriate. Then I start join in with my instrument or vocals. The session bot adapts to my input as we jam. This is a human-led generative session accompaniment.
- The other class of usage would be bot-led. I might attempt some kind of technically challenging or interesting bit that I can't quite realize. Or I may be having a bad day and struggling to get immersed in an inspiration that I've been trying to develop. The generative model might attempt to suggest a solo or effect or chord progression that helps me get back on track and in creative flux.
- Another bot-led usage could be if I'm being lazy and repetative or cliche, the bot might suggest an interesting time signature, form of syncopation or a subversive chord that fits the arrangement in a novel way or at least challenges me to not be lame. Like a coach (or badass big sister).
- I want to be able to analyze existing music compositions and other media and separate them into isolated tracks. This includes my own raw recordings of sound.
- I want to be able to store a bank of various samples from these tracks.
- I want to be able to generate high-quality emulations of these samples.
- I want to model my response to the experience of listening to music.
- I want to model my style of expressing myself musically.
- I want to generate predictive expressions.