TOOLORA / Local reading edition
GPT-SoVITS
Synthesize speech from a short reference or prepare data to fine-tune a voice; the WebUI includes vocal separation, slicing, recognition and text labeling.
Overview
GPT-SoVITS combines few-shot voice adaptation and TTS in a WebUI. Use reference speech directly or prepare a dataset to train GPT/SoVITS models. Tools cover vocal separation, automatic slicing, Chinese ASR and text labeling; documented inference languages include Chinese, English, Japanese, Korean and Cantonese. Model generations differ in resource needs. Code uses MIT; selected weights and external components require separate license checks.
Official sources and references
- Official websitehttps://github.com/RVC-Boss/GPT-SoVITS
Descriptions summarize public information; consult official websites for features, pricing and regional availability.