0.2.5

• add option to generate continuation from input audio
• upgrade to pytorch 2.0 + torch.compile compatibility
• minor fixes and formatting

0.2.4

• allow for fine tuning base models
• release pretrained checkpoints!

0.2.3

• support for new clap checkpoints, see the CLAP repo
• more customizability in clap config
• update clap code

0.2.2

• collect all acoustic tokens before generating wave, instead of generating wave section by section
• make skipping ignored files O(n)
• fix bugs with saving preprocessed samples
• sliding semantic window inference for longer outputs
• add option to filter out certain genres from fma

0.2.1

• support preprocessing longer clap windows
• add a max audio length and random crop option to preprocessor
• release some results in discord!

0.2.0

• Revamp data preprocessing. See #9

0.1.2

• remove dependency on audiolm-pytorch since we have diverged from the architecture
• quality of life things

0.1.1

• Option to provide the output feature rate for Hubert and Encodec in the config (here). This will make it easier to test out MERT v1 (which has a 75 hz feature rate vs 50 hz for MERT v0), and sets up for some upcoming preprocessing changes
• Add config for full size model with small context
• Add rvq settings to model config
• Add a script to test semantic tokens
• Fix some bugs

0.1.0

• update clap
• multi-gpu training now works for all scripts

0.0.3

option to preprocess tokens for semantic, coarse and fine stages