ByteDance launches native full-duplex speech large model Seeduplex.

On April 9th, ByteDance announced the launch of the native full-duplex speech model Seeduplex. Compared to the previous generation half-duplex bean end-to-end speech model, Seeduplex is based on a new framework design of "listening while speaking", greatly improving the naturalness and smoothness of interactive experience. Currently, Seeduplex has been fully rolled out in the Bean App, providing continuous high-quality real-time speech interaction experience for hundreds of millions of users.