The Alibaba Tongyi Laboratory voice team announced that its open-source voice generation model CosyVoice has been upgraded to version 2.0. This upgrade marks significant progress in voice generation technology in terms of accuracy, stability, and natural experience. CosyVoice 2.0 achieves bidirectional streaming voice synthesis through the integration of offline and streaming modeling techniques, allowing for a first-package synthesis delay of up to 150ms, thereby significantly improving the responsiveness of voice synthesis.

Shenzhen Riecho Technology Co., Ltd. recently issued a statement confirming that the audio portion of the audiovisual materials circulated on September 16 was cloned by the suspect using its self-developed Reecho AI voice generation model platform. The audio is derived from a live broadcast segment by Lu. The company has initiated an internal review mechanism to strengthen user identity authentication and audio generation security management to ensure the lawful and compliant use of the technology. Reecho stated that as a startup team established less than a year ago, we are committed to frontier technologies in audio and anthropomorphic artificial intelligence.

An innovative project called ChatTTS-OpenVoice has garnered attention on the internet by merging ChatTTS with OpenVoice technology, capable of generating personalized and natural-sounding voices based on a 10-second reference audio uploaded by users. This technology is suitable for dubbing tools and voice assistants, simplifying the upload process to provide users with a customized voice experience. Since its launch, the project has received widespread acclaim from developers and users alike, creating a buzz on social media. The project's leader stated that it aims to offer users a more personalized and natural voice.