MusicGen-Looper: Generate fixed-bpm loops from text prompts
1–10 of 25 posts
Re: MusicGen-Looper: Generate fixed-bpm loops from text prompts
#2Re: MusicGen-Looper: Generate fixed-bpm loops from text prompts
#3How is this possible? I thought deep learning models struggle with approximating periodic functions like sin.
"We introduce a proper inductive bias of periodicity to the generator by applying a recently proposed periodic activation called Snake function (Liu et al., 2020), defined as fα(x) = x + 1 α sin2 (αx), where α is a trainable parameter that controls the frequency of the periodic component of the signal and larger α gives higher frequency. The use of sin2 (x) ensures monotonicity and renders it amenable to easy optimization. Liu et al. (2020) demonstrates this periodic activation exhibits an improved extrapolation capability for temperature and financial data prediction."
Re: MusicGen-Looper: Generate fixed-bpm loops from text prompts
#4How is this possible? I thought deep learning models struggle with approximating periodic functions like sin.
MusicGen is an LLM on top of EnCodec tokens, instead of working directly with audio. EnCodec is neural audio compression algorithm that encodes audio as tokens from a codebook. It's a really clever trick!
Re: MusicGen-Looper: Generate fixed-bpm loops from text prompts
#5Re: MusicGen-Looper: Generate fixed-bpm loops from text prompts
#6Just seems like a fundamentally different problem than a photograph or painting.
Re: MusicGen-Looper: Generate fixed-bpm loops from text prompts
#7Re: MusicGen-Looper: Generate fixed-bpm loops from text prompts
#8In this day and age, curiosity is not worth the risk.