Chen, Jing

Research Professor

Research Interests: Auditory information processing

Office Phone: 86-10-6275 6824

Email: janechenjing@pku.edu.cn

Chen, Jing is a research professor in the Department of Machin Intelligence, School of EECS. She obtained her B.Sc. from Harbin Engineering University in 2001, and Ph.D. from Peking University in 2009 respectively. She worked as a post-doctor in the University of Cambridge from 2009 to 2012, and joined in Peking University as a research professor since 2013. Her research interests include mechanisms of auditory perception and speech perception, auditory computational models, speech signal processing, and hearing compensation technologies.

Dr. Chen has published more than 20 research papers, including journal papers on Journal of Acoustic Society of America, Speech Communication, PLOSONE, and conference papers on ICASSP and Interspeech. She was a recipient of Newton International Fellowship by Royal Society and Royal Engineering Academy, UK in 2009, and Hundred Talents Programme of Peking University, China in 2013,.

Dr. Chen is in charge of or has finished more than ten research projects including NSFC, 985 programs, SONOVA project, etc. Her research achievements are summarized as follows:

1) Computational models of speech intelligibility: The speech intelligibility index is one of the widely-used objective assessments of speech intelligibility, in which the frequency importance function is the key component, deciding the relative contribution to speech recognition made by different frequency bands. In our work, the FIF based on monosyllabic words of Mandarin Chinese was made and evaluated. The acoustic features determining the intelligibility of speech in interfering sounds were studied, including fundamental frequency contours, context tones, temporal envelopes, and new computational models were developed based on these results.

2) Algorithms of speech enhancement: Most information in speech is carried by changes in the spectrum over time. Such changes may be less audible to hearing-impaired people than to people with normal hearing, due to the reduced frequency selectivity of the former. In our studies, a form of signal processing that enhanced spectral changes over time was developed and evaluated by measuring speech intelligibility and clarity preferences for hearing-impaired listeners. The processing led to significant improvements in the intelligibility of speech in different background noises.

Technologies for hearing compensation: An important criterion to decide whether a patient is a suitable candidate for a cochlear implant is the functional integrity of the auditory neural pathways. We firstly conducted the study of electrically-evoked frequency following responses (EFFRs), and built up an experimental platform to testify the new methods. The initial results based on cochlear-impaired animal models confirm the efficiency of the methods, and further work would be cooperated with clinic researchers.