MuseTalk

⭐ 5.9k Apache-2.0 Python 2.1.0

Produced by Tencent Music, featuring 30fps+ real-time lip sync and supporting multi-language live streaming in Chinese, English, and Japanese.

📋 Info

GitHub Stars⭐ 5.9k Stars
LicenseApache-2.0
LanguagePython
Version2.1.0
Updated2026-05-25

📖 Overview

MuseTalk is a real-time lip-syncing tool developed by Tencent Music’s LyraLab (5.9k Stars). It can achieve real-time inference at 30fps+ on V100 GPUs, making it suitable for live streaming and real-time conversation scenarios. It supports multiple languages including Chinese, English, and Japanese, offering high accuracy. The model is lightweight with efficient inference performance. The community provides a variety of OBS and WebRTC streaming solutions, making it ideal for applications such as live virtual hosts and real-time digital avatars.

✨ Features

  • Real-time lip synchronization at 30fps+
  • Multi-language support for Chinese, English, and Japanese.
  • It can be used in conjunction with MuseV for full-body animation.
  • The model is lightweight, enabling efficient inference.
  • Community OBS/WebRTC streaming solutions

Advertisement

🚀 Quick Start

pip install torch==2.0.1 torchvision==0.15.2 torchaudio==2.0.2 --index-url https://download.pytorch.org/whl/cu118
pip install -r requirements.txt

🔗 Related Tools