Releases: gfdb/wav2aug
Releases · gfdb/wav2aug
v0.0.2
v0.0.1
This is an initial release for the general purpose task-agnostic speech augmentation policy, wav2aug. The policy applies two waveform augmentations in sequence to each waveform.
The release contains:
- implementations for nine audio augmentations that use our general purpose hyperparameters
- an implementation for wav2aug, which randomly samples two augmentations to apply each iteration
- some basic data loading utils to download, cache, and use data needed for augmentation (pointsource noises)
more optimizations to come!