BQ-Wang0511/TalkLikeYou-Streaming-RealTime-Demo

English | 简体中文 This browser demo generates talking-head video in real time. Animate an image using uploaded audio or a live microphone, dub a source video while preserving its pose and timing, or connect a chat agent for interactive speech-driven animation. Preset habit IDs let you switch between speaking styles without a reference video. On an NVIDIA RTX 3090, the demo achieves 30+ FPS generation throughput while using less than 4 GB of GPU memory, with high-clarity facial details. [](https://bq-wang0511.github.io/TalkLikeYou/) [](https://arxiv.org/abs/2610.06658) [](https://huggingface.co/doubi-killer/TalkLikeYou) [](https://github.com/BQ-Wang0511/TalkLikeYou) The source release includes the inference code, portrait runtime, and frontend. Model checkpoints are installed separately under the local checkpoints/ directory; runtime code never reads files from another repository. [](https://bq-wang0511.github.io/TalkLikeYou/static/videos/realtime-demo.mp4) Watch the demo video The interface provides three modes: All three modes share cached avatar preprocessing and the same streaming renderer. Audio and Chat agent accept image avatars; Video dubbing accepts video avatars and does not require…