GPT-SoVITS
TTS/cloning trainable on 1 minute of audio — a legendary Chinese-community project
A TTS/cloning solution trainable on just 1 minute of voice data; a legendary project in the Chinese community.
Text-to-speech, speech-to-text, and voice cloning, open source.
Key features
- Few-shot voice conversion and TTS WebUI
- Zero-shot TTS from a 5-second vocal sample
- Few-shot fine-tuning with 1 minute of training data
- Cross-lingual inference: English, Japanese, Korean, Cantonese, Chinese
- WebUI tools: vocal separation, dataset segmentation, ASR labeling
Good for
- Cloning voices with minimal training audio
- Beginners building TTS datasets and models via WebUI
Stack & tags
PythonPyTorchttsvoice-cloningfew-shotvoice-conversionwebuicross-lingualopen-weights
Quick start
conda create -n GPTSoVits python=3.10
conda activate GPTSoVits
bash install.sh --device <CU126|CU128|ROCM|CPU> --source <HF|HF-Mirror|ModelScope>From the project README on GitHub.
Similar open-source projects
Whisper
★ 109,795OpenAI's open-source speech recognition model — the benchmark for multilingual STT
Real-Time-Voice-Cloning
★ 60,133The classic 5-second voice cloning project — the "founding father" of voice cloning