# chrisdonahue/wavegan

WaveGAN: Learn to synthesize raw audio with generative adversarial networks

Repository: https://github.com/chrisdonahue/wavegan
Canonical: https://ross.abutalabs.com/products/wavegan
Language: Python
License: MIT
License Family: permissive
Last push: 2022-11-27T22:29:32+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3127, "days_push": 1375, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1383, forks 280 (observed 2026-08-28T04:04:34.568751+00:00)

## What it is
Official TensorFlow implementation of WaveGAN, a generative adversarial network that learns to synthesize raw audio waveforms from examples. It also includes SpecGAN, a spectrogram-based GAN alternative, and supports training on arbitrary audio files up to 4 seconds at 16kHz.

## Use cases
- generate realistic sound effects with a GAN
- synthesize speech or bird vocalizations from raw audio
- train a generative model on a folder of WAV or MP3 files
- reproduce the WaveGAN paper results on the SC09 dataset
- compare waveform GANs against spectrogram-based GANs for audio generation

## When to choose
- you want to generate short raw audio clips with a GAN
- you need the reference implementation of the WaveGAN paper
- you have a GPU and can work with TensorFlow 1.x

## When to avoid
- you need modern TensorFlow 2.x or PyTorch support
- you need to generate audio longer than 4 seconds
- you want a production-ready audio synthesis pipeline rather than research code

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, audio-processing
- domain: machine-learning, deep-learning
- platform: python
- tags: gan, audio-generation, tensorflow, waveform-synthesis, research-code, specgan, audio, gpu, linux

## Member repositories
- chrisdonahue/wavegan (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:34.568751+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:40:02.094182+00:00, confidence not recorded.
  - readme: https://github.com/chrisdonahue/wavegan (fetched 2026-08-28T04:04:34.568751+00:00, sha eeb3e3a0963e)
- Data as of 2026-08-30T08:39:29.467469+00:00.