%20--%3e%3cdefs%3e%3cstyle%3e%20.st0%20{%20fill:%20%23061b40;%20}%20.st1%20{%20fill:%20%23306af1;%20}%20.st2%20{%20fill:%20%235ce5cf;%20}%20%3c/style%3e%3c/defs%3e%3cg%3e%3cpath%20class='st0'%20d='M55,10.5h9v3h-9v9h12V7.5h-12v3ZM64,19.5h-6v-3h6v3Z'/%3e%3cpolygon%20class='st0'%20points='69%2016.5%2078%2016.5%2078%2019.5%2069%2019.5%2069%2022.5%2081%2022.5%2081%2013.5%2072%2013.5%2072%2010.5%2081%2010.5%2081%207.5%2069%207.5%2069%2016.5'/%3e%3cpolygon%20class='st0'%20points='95%2010.5%2095%207.5%2083%207.5%2083%2022.5%2095%2022.5%2095%2019.5%2086%2019.5%2086%2016.5%2095%2016.5%2095%2013.5%2086%2013.5%2086%2010.5%2095%2010.5'/%3e%3cpath%20class='st0'%20d='M40,1.5v21h11.6l1.4-1.4v-7.6h0c0,0-1.4-1.5-1.4-1.5l1.4-1.4V2.9l-1.4-1.4h-11.6ZM50,19.5h-7v-6h7v6ZM50,10.5h-7v-6h7v6Z'/%3e%3c/g%3e%3cpath%20class='st1'%20d='M23.1,24L14.7,4.8l-4.9,11.2h3.8l-1.8,4H3.3L12.1,0H2C.9,0,0,.9,0,2v20c0,1.1.9,2,2,2h21.1Z'/%3e%3cpath%20class='st2'%20d='M34,0h-16.8l10.6,24h6.2c1.1,0,2-.9,2-2V2C36,.9,35.1,0,34,0ZM32.5,20h-4V4h4v16Z'/%3e%3c/svg%3e)
On April 9, the ByteDance Seed team officially launched the native full-duplex voice large model Seeduplex, marking the evolution of AI voice interaction from traditional "turn-based" to more intuitive "real-time natural interaction." As a major upgrade of the Douyin end-to-end voice model, Seeduplex's core breakthrough lies in achieving a "listen-and-speak simultaneously" synchronous processing framework, which enhances the natural feel of voice interaction at the fundamental architecture level. Currently, this technology has been fully launched on Douyin App, completing the transition of full-duplex technology from the laboratory to large-scale user deployment.
Seeduplex improves the model's ability to resist interference in complex acoustic environments through joint modeling of speech and semantics. Compared to half-duplex solutions, its misresponse rate and misinterruption rate have been reduced by 50%. Even in situations such as navigation interference, multi-person overlap, or environmental noise, it can accurately identify the main user's intent and integrate environmental information.
In terms of dialogue rhythm control, the model introduces dynamic stopping technology, reducing stopping delay by about 250ms and decreasing the proportion of interruptions by 40%, enabling it to sensitively distinguish between users' "thinking pauses" and "end of conversation." On the engineering side, the team solved the lag issues under high concurrency through speculative sampling and quantization optimization, ensuring ultra-low latency while improving call satisfaction by an absolute value of 8.34%.
The implementation of Seeduplex is not only a leap in interaction efficiency but also signals the future trend of integrated perception, thinking, and execution. With the introduction of subsequent visual modalities, voice assistants will evolve into a multidimensional collaborative general intelligent agent capable of "listening, watching, thinking, and speaking," reshaping industry standards for smart hardware and multimodal interaction.
Project page:
https://seed.bytedance.com/seeduplex