fad-pytorch 0.0.1

Frechet Audio Distance evaluation in PyTorch

fad_pytorch

Original FAD paper (PDF)

Install

Work in progress. If you’re just finding this repo on GitHub, it may not be ready yet.

pip install fad_pytorch

About

(Intended) Features:

• runs in parallel on multiple GPUs
• favors 48kHz sample rates
• can use CLAP embeddings
• favors ops in PyTorch instead of numpy
• allows dataset access via WebDataset (over s3://)
• operates on CPU, CUDA, or MPS

This is designed to be run as 3 command-line scripts in succession:

1. fad_gen.py: produces directories of real & fake audio
2. fad_embed.py: produces directories of embeddings of real & fake audio
3. fad_score.py: reads the embeddings & generates FAD score, for real (“$r$”) and fake (“$f$”):

$$ FAD = || \mu_r - \mu_f ||^2 + tr\left(\Sigma_r + \Sigma_f - 2 \sqrt{\Sigma_r \Sigma_f}\right)$$

Related Repos

There are [several] others, but this one is mine. These repos didn’t have all the features I wanted, but I used them for inspiration:

• https://github.com/gudgud96/frechet-audio-distance
• https://github.com/google-research/google-research/tree/master/frechet_audio_distance: Goes with Original FAD paper
• https://github.com/AndreevP/speech_distances