Current research topics
- Statistical machine intelligence (particularly with probabilistic graphical models and deep learning)
- Speech recognition and understanding
- Source separation
- Speaker recognition
- Natural language processing
- Audio indexing and search
- Audio fingerprinting
- Music processing (e.g. query by humming)
- Microphone array
Past and current research projects
- NSFC 61976122 (PI): Integration of Array Perception and Speech Recognition for Far-field Sound Sources, 2020-2023.
China Electric Power Research Institute (Co-PI), Constructing domain knowledge graph and building intelligent dialog system for customer service, 2019-2020.
China Ministry of Education and China Mobile Joint Funds (PI), Customer service dialog system based on deep reinforcement learning and knowledge graph, 2018-2019.
- NSFC 61473168 (PI): Single-channel Speech Separation Based on Probabilistic Acoustic Tube Model, 2015-2018.
- Tsinghua University Initiative (PI): Brain-inspired computer audition, 2015.
- Toshiba funding (PI): ASR, 2012-2016.
- Tsinghua University Initiative (PI): Study of cognition about uncertainty in speech and language phenomena, 2012-2015.
- NSFC 61075020 (PI): Robust speech recognition with Bayesian voice modeling and adaptive noise compensation, 2011-2013.
- IBM China Research Lab funding (PI): Language Assessment for CET Spoken English Test, 2008-2009.
- China 863 2006AA01Z149 (co-PI): High performance content-based speech search, 2007-2009.
- Panasonic funding (PI): Discriminative training for robust speech recognition, 2007-2008.
- Intel China Research Center funding (PI): Person Indexing and Retrieval for movies and TVs - audio part, 2006-2008.
- Intel China Research Center funding (PI): Multimodal Soccer Video Analysis - audio part, 2006-2008.
- NSFC 60402029 (PI): Research on Refined-Structural Acoustic Modeling for Speech Recognition Based on Bayesian Networks, 2005-2007.
- Department funding (PI): Statistical methods based on Bayesian networks with selected applications, 2004-2006.