You are here

Research description Qiang Huang

In current speech recognition system, recognition is always dependent on a global language model, which is inflexible to match the quick change of information in dialogue. To handle this problem, the dynamic adaptation to a language model is thought to be an effective way. Qiang Huang will be working on tracking the change of semantic/topic/state information by combining speech with language and other context information. The related techniques developed on dynamic adaptation for language modelling will be applied to broadcast speech recognition. Qiang Huang will be also working on the development of an interactive dialogue system by integrating the state-of-art techniques in speech, language and multimodal information processing.