Westlake Heart Intelligence has announced the open-source release of its developed Westlake-Omni model. As the world's first open-source Chinese sentiment end-to-end speech interaction large model, Westlake-Omni employs a discrete representation method to unify text and speech modalities, with a particular focus on real-time performance, enabling rapid responses and nearly zero latency experiences.
The model boasts exceptional emotional understanding and expression capabilities, capable of generating clear, natural, and expressive Chinese speech. This capability is得益于 its deep training on a high-quality Chinese emotional speech dataset, allowing the model not only to comprehend complex emotions in the Chinese context but also to make interactions more human-like.
Westlake Heart Intelligence hopes that by open-sourcing the Westlake-Omni model, it will encourage more developers to participate in the development of Chinese emotional speech interaction technology, jointly promoting the advancement and application of the technology in this field.
Project Link:https://github.com/xinchen-ai/Westlake-Omni