Asia Tech Wire (Oct 23) -- Xu Tan, a former Microsoft senior researcher, joined Chinese AI startup Moonshot AI in August, focusing on developing end-to-end speech models, according to LatePost.

Moonshot AI's entire multimodal research began last October, the report said. And sources close to the company said that a team of around 10 people is currently working on video models.

Tan served as a principal researcher and research manager at Microsoft Research Asia (MSRA), focusing on generative AI and speech/audio/video content generation. Several of his language, speech, music, and video generation achievements have been applied to Azure, Bing, and other Microsoft products and services.

According to the report, one of Tan's main goals in joining Moonshot AI could be to help the AI startup create a GPT-4o-like voice experience.

OpenAI announced GPT-4o in May this year, a multimodal large model based on end-to-end speech technology that features lower latency and the ability to interrupt at any time.