jik876/hifi-gan
HiFi-GAN: Generative Adversarial Networks for Efficient and High Fidelity Speech Synthesis
Uses multi-scale discriminators with periodic pattern modeling to synthesize 22.05 kHz waveforms directly from mel-spectrograms, achieving 167× real-time speed on GPU and 13.4× on CPU. Supports both mel-spectrogram inversion and end-to-end synthesis when fine-tuned with Tacotron2 or other acoustic models, with universal pretrained weights available for transfer learning across datasets.
2,328 stars. No commits in the last 6 months.
Stars
2,328
Forks
552
Language
Python
License
MIT
Category
Last pushed
Jul 27, 2024
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/voice-ai/jik876/hifi-gan"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related tools
yeyupiaoling/MASR
Pytorch实现的流式与非流式的自动语音识别框架，同时兼容在线和离线识别，目前支持Conformer、Squeezeformer、DeepSpeech2模型，支持多种数据增强方法。
coqui-ai/TTS
🐸💬 - a deep learning toolkit for Text-to-Speech, battle-tested in research and production
shivammehta25/Matcha-TTS
[ICASSP 2024] 🍵 Matcha-TTS: A fast TTS architecture with conditional flow matching
netease-youdao/EmotiVoice
EmotiVoice 😊: a Multi-Voice and Prompt-Controlled TTS Engine
gabrielmittag/NISQA
NISQA - Non-Intrusive Speech Quality and TTS Naturalness Assessment