|RESEARCH AREAS||PROJECTS||PUBLICATIONS||TECHNOLOGY TRANSFER|
Human Language Technology
Speech is the most direct and natural means for human-machine communication. Our research focuses on constructing a platform for automatic speech recognition which can handle medium-vocabulary, speaker-independent, continuous and spontaneous speech, in noisy environment. Major research projects include time-scale modification of speech waveform for data compression; robust recognition of speech, dealing with accent and channel differences; acoustic modeling of Chinese and mixed language speech based on acoustic modeling of subword units and tonal recognition; understanding of ungrammatical speech using segmentation and categorization; as well as dialog management to handle interactive speech. As part of the joint EEE/CS Human Language Technology Centre, we collaborate with Computer Science faculties to build a spoken language translation system.