fad-pytorch 0.0.2

Frechet Audio Distance evaluation in PyTorch

fad_pytorch

Original FAD paper (PDF)

Install

pip install fad_pytorch

About

(Intended) Features:

• runs in parallel on multiple GPUs
• supports 48kHz sample rates and stereo when possible
• supports CLAP embeddings, in addition to VGGish and PANN
• favors ops in PyTorch instead of numpy
• allows dataset access via WebDataset (over s3://)
• runs on CPU, CUDA, or MPS

This is designed to be run as 3 command-line scripts in succession. The latter 2 (fad_embed and fad_score) are probably what most people will want:

1. fad_gen: produces directories of real & fake audio
2. fad_embed <real_audio_dir> <fake_audio_dir>: produces directories of embeddings of real & fake audio
3. fad_score <real_emb_dir> <fake_emb_dir>: 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