Volcano Engine launches SeedRealtime 1.0 audio-video call model API
Volcano Engine has launched the SeedRealtime 1.0 real-time audio-video call model API, according to 36Kr and IT之家. The model combines sound, visuals, timing and expression in one end-to-end system for continuous “see, hear and speak” interaction. IT之家 reports that it reduces the cumulative delay and information loss associated with chaining separate ASR, VLM and TTS modules, and says its audio-video conversation rhythm problems were halved compared with cascaded models. The supplied reports do not include independent verification of these performance comparisons.
SeedRealtime 1.0 offers developers a single API for jointly processing speech, visual scenes and timing information in one real-time interaction loop, rather than linking separate perception and speech modules.
火山引擎上线豆包实时音视频通话模型API服务-36氪
登录
搜索