The University of Tokyo
Google Scholar · Lab Profile · Hugging Face
I work on computer graphics and generative models, with a focus on human motion, speech synthesis, and interactive digital humans. I am also interested in real-time rendering, physically based simulation, and building expressive virtual worlds.
Efficient and Path Controllable Streaming Motion Generation
Streaming motion generation with KV caching, geometry-aware supervision, and root-path control.
Paper · Project · Code · Models · Live Demo
Tailored Diffusion Forcing for Streaming Motion Generation
Real-time, continuous human motion generation driven by changing text prompts.
Paper · Project · Code · Models
Kinetic-Optimal Scheduling with Moment Correction for Metric-Induced Discrete Flow Matching in Zero-Shot Text-to-Speech
Zero-shot speech synthesis with kinetic-optimal scheduling and moment correction.
Shallow Flow Matching for Coarse-to-Fine Text-to-Speech Synthesis
A coarse-to-fine flow matching method for speech synthesis.
Towards Interactive Intelligence for Digital Humans
Interactive digital humans combining reasoning, speech, facial animation, body motion, and rendering.
Automatic Texture Compression for Facial Blendshapes
Compact dynamic facial textures for blendshape animation in real-time applications.
See Google Scholar for my publication list.