According to CoinMeta, ByteDance has released DouBao's real-time visual capabilities today. Users can watch, listen, and speak simultaneously through DouBao app, and the system can also proactively remind users. Compared to the previous generation seeduplex, the new version adds real-time visual processing capabilities that can identify speakers in multi-person and noisy environments, and understand context by combining facial features, voice, and gestures. ByteDance claims that this feature reduces issues such as interrupting others, slow reactions, and false triggers due to noise by about half compared to traditional splicing solutions.