SUSTech Audio Intelligence Lab

This is SUSTech Audio Intelligence Lab (SAIL, 南科大音频智能实验室) directed by Prof. Zhong-Qiu Wang in the Department of Computer Science and Engineering at Southern University of Science and Technology (SUSTech) in Shenzhen, China.

We are interested in broad speech/audio signal processing and artificial intelligence problems, aiming at building machine listening systems that can robustly perceive and understand speech/audio in reverberant environments with multiple concurrent sound sources. Our current research focuses on deep learning based approaches for speech enhancement, speaker separation, speech dereverberation and robust automatic speech recognition based on a single microphone or an array of microphones, targeting at solving the cocktail party problem.

招生信息
• 招收1名博士学生（广东省工程硕博专项，2026年秋季入学）
• 招收1名博士学生（快响计划，2026年秋季入学）

News

[03/2026] - Our paper "The SUSTech AILab System Description for CHiME-9 MCoRec Challenge" is accepted by ICASSP 2026 Workshop on HSCMA and CHiME.
[01/2026] - 7 papers are accepted by ICASSP 2026.
[11/2025] - Prof. Wang is invited to serve as an Action Editor for the Neural Networks journal.
[11/2025] - Our paper "Listen to Extract: Onset-Prompted Target Speaker Extraction" is accepted by TASLPRO.
[11/2025] - Our paper "Recent Trends in Distant Conversational Speech Recognition: A Review of CHiME-7 and 8 DASR Challenges" is accepted by Computer Speech & Language.
[09/2025] - Our paper "ctPuLSE: Close-Talk, and Pseudo-Label Based Far-Field, Speech Enhancement" is accepted by JASA.
[09/2025] - Prof. Wang is listed in "World's Top 2% Scientists List - Single-Year Impact" (全球前2%顶尖科学家年度榜单), 2025.
[08/2025] - Our paper "Evolutionary Prompt Design for LLM-Based Post-ASR Error Correction" is accepted by SiPS 2025.
[07/2025] - Our paper "Unsupervised Multi-Channel Speech Dereverberation via Diffusion" is accepted by WASPAA 2025.
[07/2025] - We rank third place in "DCASE2025 Challenge Task 4 - Spatial Semantic Segmentation of Sound Scenes". See our technical report for our solutions.
[06/2025] - Our paper "Unsupervised Multi-Channel Speech Dereverberation via Diffusion" is accepted by ICML Workshop on Machine Learning for Audio 2025.
[05/2025] - Our papers "ARiSE: Auto-Regressive Multi-Channel Speech Enhancement", "Multi-Channel Acoustic Echo Cancellation Based on Direction-of-Arrival Estimation", and "AuralNet: Hierarchical Attention-based 3D Binaural Localization of Overlapping Speakers" are accepted by Interspeech 2025.
[05/2025] - Our paper "Unsupervised Blind Speech Separation with A Diffusion Prior" is accepted by ICML 2025.
[04/2025] - Our paper "An End-to-End Integration of Speech Separation and Recognition with Self-Supervised Learning Representation" is accepted by Computer Speech & Language.
[03/2025] - Our paper "SuperM2M: Supervised and Mixture-to-Mixture Co-Learning for Speech Enhancement and Noise-Robust ASR" is accepted by Neural Networks.
[12/2024] - Our paper "30+ Years of Source Separation Research: Achievements and Future Challenges" is accepted by ICASSP 2025.
[08/2024] - Our paper "USDnet: Unsupervised Speech Dereverberation via Neural Forward Filtering" is accepted by IEEE/ACM TASLP.