hi-paris/wavlm-vocoder-french
WavLM-to-Audio neural vocoder for French speech reconstruction — layer ablation study and adversarial supervision as a foundation for continuous voice conversion (JEP 2026)
This tool helps researchers and voice technology specialists reconstruct high-quality French speech from advanced acoustic representations. It takes a compressed representation of French speech (WavLM features) and converts it back into natural-sounding audio. This is particularly useful for scientists working on foundational speech technologies.
Use this if you need to generate realistic French speech audio from WavLM features, especially as a building block for continuous voice conversion projects.
Not ideal if you're looking for a direct voice cloning solution for end-users or need to process languages other than French.
Stars
18
Forks
—
Language
Python
License
MIT
Category
Last pushed
Mar 25, 2026
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/voice-ai/hi-paris/wavlm-vocoder-french"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
kan-bayashi/ParallelWaveGAN
Unofficial Parallel WaveGAN (+ MelGAN & Multi-band MelGAN & HiFi-GAN & StyleMelGAN) with Pytorch
fatchord/WaveRNN
WaveRNN Vocoder + TTS
shangeth/wavencoder
WavEncoder is a Python library for encoding audio signals, transforms for audio augmentation,...
rishikksh20/iSTFTNet-pytorch
iSTFTNet : Fast and Lightweight Mel-spectrogram Vocoder Incorporating Inverse Short-time Fourier...
seungwonpark/melgan
MelGAN vocoder (compatible with NVIDIA/tacotron2)