ByteDance launches SeedRealtime 1.0 end-to-end realtime audio-video model API
火山引擎 · wechat · 2026-11-19
Volcano Engine launched the API for SeedRealtime 1.0, a Doubao end-to-end realtime audio-video model that unifies sound, vision, timing and expression in one model, replacing cascaded ASR+VLM+TTS pipelines. It features continuous environment awareness, proactive speech, multi-person visual grounding and noise robustness, demonstrated across six real-world scenarios; dialogue pacing issues halved vs cascaded models.
More from Multimodal
- NVIDIA open-sources MotionBricks, a real-time motion generator trained on 350K+ clips — Scobleizer · 2026-11-19
- Movie2Game-Bench tests 14 models on film-to-game adaptation; playable sequences score just 37.9/100 — Jialing Xie · 2026-11-19
- Meshy releases UltraTex for physically plausible PBR texture generation — meshy-ai-team · 2026-11-19
- Redditor plans €300 first test of programmable AI video editing workflow — yeah280 · 2026-11-19
- Lag correction cuts SiT-XL/2 FID from 28.0 to 8.7 on ImageNet — cloneofsimo · 2026-11-19
- AI-generated 'valley where stars flow like water' short stuns Reddit — the-quiet-unreal · 2026-11-19