I will join the SUSTech Audio Intelligence Lab (SAIL) as a Ph.D. student, under the supervision of Prof. Zhong-Qiu Wang.
My doctoral research will focus on speech processing in cocktail-party scenarios.
Previously, I was a Master’s student at the Technical University of Denmark, majoring in Electrical Engineering.
During my M.Sc. studies, I was jointly supervised by Prof. Jinqiu Sang from ECNU and Prof. Jens Hjortkjær from DTU.
My research interests lie in speech and auditory signal processing, with a particular focus on the Cocktail-Party Problem, including auditory attention decoding, speech separation, denoising, and target speaker extraction. I am also interested in extending these directions toward SpeechLLMs.
I am broadly interested in audio signal processing, including perceptual and data-driven models for automotive audio systems. This interest is motivated by my long-standing passion for music. I welcome academic and industrial collaborations via email.
🔥 News
- 2026: 🎓 Starting Ph.D. at Southern University of Science and Technology (SUSTech), Audio Intelligence Lab (SAIL).
- 2025.04: 🎉🎉 Started an audio algorithm research internship at GoerDynamics – Dynaudio Lab.
- 2025.04: 📄 One paper accepted by ICIC 2025.
- 2023.09: 🎓 Started my Master’s degree at Technical University of Denmark.
💻 Internships
- 2025.04 - 2025.07, audio algorithm research, GoerDynamics-Dynaudio Lab, China. During internship, I worked on end-to-end parametric equalizer prediction models for automotive audio systems, aiming to simplify and automate the manual sound tuning process. This work was carried out under the supervision of Technical Lead Jörg Lichtenstein.
📝 Publications
Multi-Scale Physiologically-Motivated Alignment for Auditory Attention Decoding
Yuxuan Ma, etc.
We propose a method for the auditory attention decoding (AAD) match–mismatch paradigm that explicitly models the temporal latency between EEG and speech representations, enabling more accurate time alignment across modalities.
[Paper under review]
Ear-Acoustic Authentication with Personalized Music: Binaural Robustness against Playback Attacks
Tongxi Chen, Yuxuan Ma, etc.
We explore the feasibility of ear-canal–based acoustic probing signals for continuous user authentication, and evaluate the system’s robustness against imitation attacks.
[Paper under review]
HERMES: Heterogeneous Mixture of Experts Based on Segments for Auditory Attention Decoding
Yuxuan Ma, Jun Xue, Jinqiu Sang
We investigate learning-based approaches for bridging the representation gap between EEG and speech modalities, aiming to improve cross-modal alignment and matching.
📖 Education
- 2026 -, Ph.D. in Computer Science, Southern University of Science and Technology, Shenzhen, China.
- 2023.09 - 2025.12, M.Sc. in Electrical Engineering, Technical University of Denmark, Lyngby, Denmark.
- 2024.09 - 2025.01, M.Sc. Exchange, The Hong Kong University of Science and Technology, Hong Kong, China.
- 2019.09 - 2023.06, B.Eng. in Electronic Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing, China.
🎖 Honors and Awards
- 2023 Outstanding Graduate
- 2020~2022 Academic Scholarship
- 2020 Merit Student Award