Zhengrui Ma

FastLongSpeech: Enhancing Large Speech-Language Models for Efficient Long-Speech Processing

Efficient Speech Language Modeling via Energy Distance in Continuous Latent Space

LLaMA-Omni: Seamless Speech Interaction with Large Language Models

Overcoming Non-monotonicity in Transducer-based Streaming Generation

Learning Monotonic Attention in Transducer for Streaming Generation