Alibaba Tongyi Bailing launches a new version of its voice model

Ali Tongyi announced the open-source release of Fun-CosyVoice3-0.5B, which provides zero-shot voice cloning capabilities. You only need to provide a reference audio of more than 3 seconds to replicate its tone and synthesize new speech. It also supports local deployment and secondary development. In addition, Tongyi launched a lightweight version, the Fun-ASR-Nano model, with a total parameter count compressed to 0.8B, resulting in lower inference costs. It is now open-source and supports local deployment and customized fine-tuning. (Interface)