Sesame Inc. has introduced its latest speech synthesis model, CSM, leveraging an end-to-end Transformer multimodal learning architecture. This advanced model excels at understanding context and producing natural, emotionally charged speech with voice effects that mirror those of a real person. CSM boasts real-time speech generation capabilities, enabling users to fine-tune parameters such as tone, intonation, and more, showcasing remarkable flexibility. While primarily trained in English, the model currently does not support Chinese, yet it holds promising potential for future multilingual integration.
