5 papers
MeanFlowSE: one-step generative speech enhancement via conditional mean flow
Duojia Li, Shenghui Lu, Hongchen Pan +3
Multistep inference is a bottleneck for real-time generative speech enhancement because flow- and diffusion-based systems learn an instantaneous velocity field and therefore rely o…
Pseudo Labels-based Neural Speech Enhancement for the AVSR Task in the MISP-Meeting Challenge
Longjie Luo, Shenghui Lu, Lin Li +1
This paper presents our system for the MISP-Meeting Challenge Track 2. The primary difficulty lies in the dataset, which contains strong background noise, reverberation, overlappin…
A Two-Stage Hierarchical Deep Filtering Framework for Real-Time Speech Enhancement
Shenghui Lu, Hukai Huang, Jinanglong Yao +3
This paper proposes a model that integrates sub-band processing and deep filtering to fully exploit information from the target time-frequency (TF) bin and its surrounding TF bins…
SlimSpeech: Lightweight and Efficient Text-to-Speech with Slim Rectified Flow
Kaidi Wang, Wenhao Guan, Shenghui Lu +3
Recently, flow matching based speech synthesis has significantly enhanced the quality of synthesized speech while reducing the number of inference steps. In this paper, we introduc…
Dynamic Language Group-Based MoE: Enhancing Code-Switching Speech Recognition with Hierarchical Routing
Hukai Huang, Shenghui Lu, Yahui Shan +5
The Mixture of Experts (MoE) model is a promising approach for handling code-switching speech recognition (CS-ASR) tasks. However, the existing CS-ASR work on MoE has yet to levera…