Jinyu Li

EmotionThinker: Prosody-Aware Reinforcement Learning for Explainable Speech Emotion Reasoning

FlexiCodec: A Dynamic Neural Audio Codec for Low Frame Rates

CoVoMix2: Advancing Zero-Shot Dialogue Generation with Fully Non-Autoregressive Flow Matching

ARLON: Boosting Diffusion Transformers with Autoregressive Models for Long Video Generation

VALL-E 2: Neural Codec Language Models are Human Parity Zero-Shot Text to Speech Synthesizers

RALL-E: Robust Codec Language Modeling with Chain-of-Thought Prompting for Text-to-Speech Synthesis