https://github.com/fishaudio/vocoder

https://github.com/fishaudio/vocoder

Last synced: 8 months ago
JSON representation

• Host: GitHub
• URL: https://github.com/fishaudio/vocoder
• Owner: fishaudio
• License: mit
• Created: 2023-06-15T00:40:01.000Z (about 3 years ago)
• Default Branch: main
• Last Pushed: 2025-06-02T18:27:41.000Z (about 1 year ago)
• Last Synced: 2025-06-05T07:25:22.846Z (about 1 year ago)
• Language: Python
• Size: 421 KB
• Stars: 108
• Watchers: 3
• Forks: 8
• Open Issues: 5
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Fish Vocoder

This repo is designed as an uniform interface for developing various vocoders.

Configs:

- [x] hifigan (baseline): HiFiGAN generator with UnivNet discriminators.

- [x] bigvgan: BigVGAN generator.

- [x] vocos: Vocos (ConvNext) generator.

- [x] refinegan: RefineGAN generator.

- [ ] firefly-gan: convnext encoder + hifigan generator.

## References

- TIMM: https://github.com/huggingface/pytorch-image-models

- BigVGAN: https://github.com/NVIDIA/BigVGAN

- Vocos: https://github.com/charactr-platform/vocos

- UnivNet: https://github.com/mindslab-ai/univnet

- ConvNext: https://github.com/facebookresearch/ConvNeXt

- HiFiGAN: https://github.com/jik876/hifi-gan

- Fish Diffusion: https://github.com/fishaudio/fish-diffusion

- RefineGAN: https://arxiv.org/abs/2111.00962

- Encodec: https://github.com/facebookresearch/encodec

- EVA-GAN: https://arxiv.org/abs/2402.00892

- ConvNext: https://github.com/facebookresearch/ConvNeXt