---

AUDIO team at Westlake University works on acoustic, audio and speech processing. Our research goal is to develop new models/algorithms that enable machine/robot to listen (auditory perception) like human beings, more specifically being able to extract valuable information in complex and dynamic acoustic environments. This will leverage advanced (microphone array) signal processing and deep learning techniques to efficiently employ temporal, spatial and spectral cues of sound recordings, with an emphasis on developing signal-prior-guided deep learning techniques. In addition, we also work on personal sound zone (personalized sound field reproduction), paying more attention on the topics that would facilitate the practical use, such as developing low-complexity and adaptive algorithms.

• Specific research topics include:
• 1) Speech/sound processing frontends: sound/speech denoising, dereverberation, separation and localization;
• 2) Speech/sound processing backends: sound/speech semi-supervised learning and unsupervised pre-training;
• 3) Acoustic virtual/augmented reality: personal sound zone.

---

---