We are the TaoLiveAIGC Team at Taobao & Tmall, Alibaba Group, building next-generation digital humans for live commerce. Our work spans speech, LLMs, video generation, and multimodal intelligence, enabling natural and real-time interaction at scale.

• Github: https://github.com/TaoLiveAIGC
• Huggingface: https://huggingface.co/TaoLiveAIGC

💻 Research Projects

🎙️ TTS

• Qwen3-TTS-TBLive: Live-Streaming Domain-adaption Built on Qwen3-TTS-1.7B Base model
  

🧠 Large Language Model

🎥 Video Generation

• 🚀TaoMate-HD: Progressive-Resolution Autoregressive Audio-Video Generation
  

• 🚀TaoMate-FVSR: Audio-Guided Facial Video Restoration
  

• 🚀TaoMate-H3: TaoMate training with MiniMax H3. Streaming video generation with 3-step LoRA.
  

  • 🤩2026/10/15: We release the first streaming FL2AV model. Have fun! ❤️
  • 🤩2026/09/08: We first migrate MiniMax H3 to TaoMate (T2AV) for streaming, with the complete streaming FL2AV and Ref2AV on the way, please stay tuned.

• TaoMate: Anchor-Guided Memory Bridging Evolving and Reference States for Real-Time Audio-Video Digital Human Generation
  

• TBDub: Production-Oriented Visual Dubbing
  

• 🚀AptAvatar: Fast and Vivid Long-Form Audio-Driven Video Generation for Production-Ready Avatars
  

• CoInteract: Spatially-Structured Co-Generation for Interactive Human-Object Video Synthesis
  

🌐 Multimodal Large Language Model

• 🚀TLive-Omni: An Omni-Modal Understanding Model for E-Commerce Live Streaming