Projects

PSLM (Parallel Speech Language Model)
Parallel generation of text and speech using LLM for low-latency spoken dialogue.
arXiv Demo
PSLM (Parallel Speech Language Model)
Nue ASR
Integrating pretrained HuBERT and GPT for automatic speech recognition.
arXiv GitHub Hugging Face
Nue ASR
CHATS (CHatty Agents Text-to-Speech)
Natural AI-to-AI conversation with spoken content control over written dialogue.
arXiv Demo
CHATS (CHatty Agents Text-to-Speech)
Koemotion
Japanese text to speech and facial keypoint with a speaker control over 2D map.
Koemotion
UniFLG (Unified Facial Landmark Generator)
Integrating audiovisual speech synthesis (text to speech and face) and speech-driven facial animation (speech to face) for multimodal interaction.
arXiv Demo
UniFLG (Unified Facial Landmark Generator)