ByteDance Launches Native Full-Duplex Voice Large Model Seeduplex
Published · Apr 9 · Thu Source · 字节跳动Seed (CN)

ByteDance Launches Native Full-Duplex Voice Large Model Seeduplex

ByteDance Seed team launches full-duplex voice large model Seeduplex, now fully available on the Doubao App. The model is based on a "listen-and-speak" framework, achieving synchronized listening and speaking, with precise anti-interference and dynamic endpointing capabilities. The false reply rate and interruption ratio are reduced by 50% and 40% respectively, and endpointing latency is reduced by 250ms. Evaluations show a 12% improvement in conversation fluency and an 8.34% increase in user satisfaction, marking the first large-scale implementation of full-duplex voice technology.

KeywordsByteDanceLaunchesNativeFull-DuplexVoiceLargeModelSeeduplex

ByteDance Seed team launches full-duplex voice large model Seeduplex, now fully available on the Doubao App. The model is based on a "listen-and-speak" framework, achieving synchronized listening and speaking, with precise anti-interference and dynamic endpointing capabilities. The false reply rate and interruption ratio are reduced by 50% and 40% respectively, and endpointing latency is reduced by 250ms. Evaluations show a 12% improvement in conversation fluency and an 8.34% increase in user satisfaction, marking the first large-scale implementation of full-duplex voice technology.

April 10, 2026 • What is Seeduplex? ByteDance's Full-Duplex Voice AI Explained A plain-English breakdown of how Seeduplex works and why full-duplex changes everything. Apr 10, 2026 …

The Seeduplex model can continuously receive and understand user-side audio, perceive the global acoustic environment where the user is located, thereby accurately judging which sounds are truly interacting with the model and which are interference. The improvement in anti-interference capability has significantly reduced Seeduplex's false reply rate and false interruption rate …

Cross-provider support for 140+ languages. ByteDance Seed Speech and iFlytek provide professional Chinese support. Register and obtain Seeduplex API Key. Free quota requires no credit card. Specify model or enable intelligent routing. SDK automatically handles authentication, retries, and failover. Enab…

April 9, 2026 • Seeduplex significantly improves endpointing performance by 8% compared to the half-duplex solution. When responding to interruptions, real human dialogue experiences occasional lag, ….

This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.