OpenVoice is an innovative voice cloning approach that has the ability to replicate a person’s voice and generate speech in multiple languages. It addresses several challenges in the field, including flexible voice style control and zero-shot cross-lingual voice cloning. Unlike previous methods, OpenVoice allows for granular control over voice styles, such as emotion, accent, rhythm, pauses, and intonation. It can also clone voices into new languages without the need for extensive training data. Additionally, OpenVoice is computationally efficient and cost-effective compared to other APIs. The source code and trained model are publicly accessible, encouraging further research in the field.
https://arxiv.org/abs/2312.01479