6 papers
SPG-Codec: Exploring the Role and Boundaries of Semantic Priors in Ultra-Low-Bitrate Neural Speech Coding
Mingyu Zhao, Zijian Lin, Kun Wei +1
Conventional neural speech codecs suffer from severe intelligibility degradation at ultra-low bitrates, where the bottleneck transitions from acoustic distortion to semantic loss.…
Hearing More with Less: Multi-Modal Retrieval-and-Selection Augmented Conversational LLM-Based ASR
Bingshen Mu, Hexin Liu, Hongfei Xue +2
Automatic Speech Recognition (ASR) aims to convert human speech content into corresponding text. In conversational scenarios, effectively utilizing context can enhance its accuracy…
Mixture of LoRA Experts with Multi-Modal and Multi-Granularity LLM Generative Error Correction for Accented Speech Recognition
Bingshen Mu, Kun Wei, Pengcheng Guo +1
Despite improvements in automatic speech recognition, performance drops with accented speech. Generative error correction (GER) leverages the linguistic knowledge of large language…
OSUM: Advancing Open Speech Understanding Models with Limited Resources in Academia
Xuelong Geng, Kun Wei, Qijie Shao +18
Large Language Models (LLMs) have made significant progress in various downstream tasks, inspiring the development of Speech Understanding Language Models (SULMs) to enable compreh…
DQ-Data2vec: Decoupling Quantization for Multilingual Speech Recognition
Qijie Shao, Linhao Dong, Kun Wei +2
Data2vec is a self-supervised learning (SSL) approach that employs a teacher-student architecture for contextual representation learning via masked prediction, demonstrating remark…
HDMoLE: Mixture of LoRA Experts with Hierarchical Routing and Dynamic Thresholds for Fine-Tuning LLM-based ASR Models
Bingshen Mu, Kun Wei, Qijie Shao +2
Recent advancements in integrating Large Language Models (LLM) with automatic speech recognition (ASR) have performed remarkably in general domains. While supervised fine-tuning (S…