Found 1 document(s)
Page 1 of 1
-
Controlling melody in audio generation normally requires training a conditional generative model on audio paired with pitch curves or MIDI notes. We instead guide an existing unconditional diffusion model, originally tra...