抄録
Acoustic modeling is one of the most important aspects of a speech recognition system. In recent years, most research has focused on probabilistic models, namely hidden Markov models. For achieving the recognition of spontaneous speech from unrestricted speakers, precise and robust acoustic modeling becomes even more important. This paper describes a framework for achieving precise and robust hidden Markov modeling, and the most recent research at ATR for refining the framework and achieving even finer acoustic models.
