The Shenzhen Association for Artificial Intelligence (SAAI) recently announced the winners of the 2023 Shenzhen Artificial Intelligence Awards. Professor Haizhou Li’s interdisciplinary team from the School of Data Science at The Chinese University of Hong Kong, Shenzhen, and the Shenzhen Research Institute of Big Data has won the Shenzhen Artificial Intelligence Natural Science Award. Professor Li, the X.Q. Deng Presidential Chair Professor at the School of Data Science and Chief Scientist at the Shenzhen Research Institute of Big Data, led the award-winning project titled "Solution to the Cocktail Party Problem – Neuromorphic Speaker Extraction Technology and Applications in Complex Acoustic Scenarios." This project involved collaborations with the School of Data Science and the School of Medicine at CUHK-Shenzhen, the Shenzhen Research Institute of Big Data, the Shenzhen University General Hospital, and Kuaishou Technology.

The third "Shenzhen Artificial Intelligence Award"  list in 2023

Source: Shenzhen Artificial Intelligence Association Official Website

Introduction of Team Members

Team Leader

Haizhou Li
SAEng Fellow
IEEE Signal Processing Society Vice President (2024-2026)
IEEE/ISCA/AAlA Fellow
X.Q. Deng Presidential Chair Professor, Executive Dean
Chief Scientist of Shenzhen Research Institute of Big Data

Research Field
Speech Information Processing, Natural Language Processing, Neuromorphic Computing, Human-Computer Interface

Biography
Professor Haizhou Li is the X.Q. Deng Presidential Chair Professor in the School of Data Science, The Chinese University of Hong Kong, Shenzhen, China. He is also an Adjunct Professor at the National University of Singapore, Singapore and a Bremen Excellence Chair Professor at the University of Bremen, Germany. Prior to joining CUHK (Shenzhen), Professor Li has taught at Nanyang Technological University and National University of Singapore (2006-2016) in Singapore, University of Eastern Finland (2009) in Finland, and University of New South Wales (2011-2016) in Australia. He was the Principal Scientist and Research Director at the Institute for Infocomm Research (2003-2016), the Agency for Science, Technology and Research, Singapore. Professor Li is an IEEE Fellow, and ISCA Fellow.

Professor Li has served as the Editor-in-Chief of IEEE-ACM Transactions on Audio Speech and Language Processing (2015-2018), Associate Editor of Computer Speech and Language (2012-2021), Springer International Journal of Social Robotics (2008-2021), and a Member of IEEE Speech and Language Processing Technical Committee (2013-2015), Awards Board (2021-2023), and Publications Board (2015-2018) of IEEE Signal Processing Society. He was the President of the International Speech Communication Association (ISCA, 2015-2017), the President of Asia Pacific Signal and Information Processing Association (APSIPA, 2015-2016), the President of the Asian Federation of Natural Language Processing (AFNLP, 2017-2018). He was the General Chair of major scientific conferences including ACL 2012, INTERSPEECH 2014, and ICASSP 2022.

Professor Li was known for his technical contributions to the theory and practice of speaker and language recognition, and to a number of award-winning speech products, such as Apple's Chinese Dictation Kits for Macintosh (1996) and Lernout & Hauspie's Speech-Pen-Keyboard Text Entry Solution for Asian languages (1999). He was the architect of a series of major technology deployments that include TELEFIQS voice-automated call centre service in Singapore Changi International Airport (2001), voiceprint engine for Lenovo A586 Smartphone (2012), and Baidu Music Search (2013).
 

Team members

Siqi Cai
Researcher, Shenzhen Research Institute of Big Data (SRIBD)

Dr. Siqi Cai is a leading young talent in the fields of brain science and neuromorphic computing. Dr. Cai has made significant innovative contributions to decoding auditory information from brain signals and neuromorphic computing methods based on auditory cognitive principles. In this project, Dr. Cai is responsible for EEG signal processing, EEG attention detection, and implementing the cocktail party solution.

Wei Tao
Director of Neurosurgery, Associate Professor, and Master’s Supervisor at Shenzhen University General Hospital
National Committee Member and Work Secretary of the Chinese Association for the Study of Pain (CASP)
Deputy Leader of the Maxillofacial Pain Study Group of the CASP

Director Wei Tao has over 20 years of clinical experience in neurosurgery, focusing on the mechanisms and neuroregulatory treatments of functional brain diseases. She has extensive experience in diagnosing and treating chronic pain, epilepsy, Parkinson’s disease, cerebral palsy, hemifacial spasm, coma arousal, and nervous system tumours. She has led and participated in numerous projects funded by the National Natural Science Foundation, the Beijing Natural Science Foundation, and the Shenzhen Science and Technology Innovation Commission. In this project, Director Tao is responsible for the medical acquisition and annotation of EEG data, and providing guidance for building interpretable models.

Renzhi Wang

Consultant, Professor, and Ph.D. Supervisor at the School of Medicine, The Chinese University of Hong Kong, Shenzhen
Director of the Neurosurgery Teaching and Research Office, Chief Physician at Peking Union Medical College Hospital
Deputy Director of the Regenerative Medicine Department (Eight-Year Programme) at Peking Union Medical College

Professor Renzhi Wang is a consultant, professor, and Ph.D. supervisor at the School of Medicine, The Chinese University of Hong Kong, Shenzhen. He is also the director of the Neurosurgery Teaching and Research Office and chief physician at Peking Union Medical College Hospital, a second-level professor, and the deputy director of the Regenerative Medicine Department (Eight-Year Programme) at Peking Union Medical College. He is a recipient of the National 70th Anniversary Commemorative Medal. In this project, Professor Wang is responsible for the analysis and interpretation of EEG signals, providing neuroscientific theoretical support for brain cognitive mechanisms.

Chenglin Xu
Audio Algorithm Expert at Kuaishou Technology

Dr. Chenglin Xu is a leading young talent in the fields of speech processing and auditory attention. He has published over 40 papers in renowned journals and conferences indexed by SCI, with over a thousand citations. He has won multiple international speech processing competitions. In this project, Dr. Xu innovatively proposed a speaker extraction technology based on human auditory attention mechanisms, offering a new solution to the "cocktail party problem" that has persisted for nearly half a century. His open-source speaker extraction technology has been widely adopted in both industry and academia.

Program Introduction

The human ear and brain together form a complete auditory system capable of selectively listening to specific sounds in noisy environments, known as the "cocktail party effect." However, current intelligent devices struggle to match human performance in this regard. The "cocktail party problem" is a classic issue in computational auditory processing, critical for tasks such as speech recognition, speech enhancement, speaker recognition, and speech separation. As intelligent devices and portable computing technologies advance, solving this problem becomes increasingly essential.

This project addresses three significant scientific questions to solve the "cocktail party problem":

a) How to effectively separate target sound sources in complex environments?

b) How to detect the speaker the brain is listening to from brain signals?

c) How to use brainwave signals to extract the target sound source?

By integrating knowledge-driven methods based on brain auditory mechanisms and signal processing with data-driven methods based on deep learning, the project aims to develop assistive listening devices with human-like auditory attention capabilities in complex scenarios.

The project proposed a method for automatically detecting target sound sources in brain auditory activities and extracting target speakers in complex acoustic scenes, leading to the successful development of a brain-controlled assistive listening device. Specifically, the project investigated a universal auditory front-end model that integrates multiple modalities and information sources. Combining neuroscience and brain cognition research, the project explored auditory front-end processing mechanisms, achieving computational interpretability. To meet the needs of different application scenarios, a multi-level decoding mechanism for the target sound source's speech characteristics, location features, and speaker identity was designed and implemented, efficiently filtering out interfering sources and background noise, providing users with clear target sound sources. Additionally, the project established the first auditory cognitive database with tonal language sound sources and proposed effective evaluation methods, providing data resources and assessment standards for researchers worldwide.
 

Brain-Controlled Intelligent Assistive Listening System
 

In summary, this project provides the theoretical foundation and application technology for brain-controlled assistive listening devices, audio front-ends for video calls, and intelligent devices. According to the "Shenzhen Future Industries Development Action Plan (2022-2025)," brain science and neuromorphic intelligence are among the eight key industries for Shenzhen’s future development, highlighting the project's significant scientific, social, and practical value.

Shenzhen Association for Artificial Intelligence，SAAI

The Shenzhen Association for Artificial Intelligence (SAAI) is a local, non-profit academic organisation legally registered with the Shenzhen Civil Affairs Bureau, guided by the Shenzhen Association for Science and Technology.
 

SAAI aims to unite and lead its members while adhering to the constitution, laws, regulations, national policies, and social ethics. It conducts activities related to artificial intelligence science and technology, promoting theoretical research, application development, and the prosperity of the AI industry in Shenzhen. The association's core tasks include academic exchange, scientific education, academic publishing, talent recommendation, academic evaluation, consultancy, technical review, and awards, contributing to the development of Shenzhen's AI industry, social progress, and security enhancement.

Source: Shenzhen Artificial Intelligence Association Official Website