CONTINUOUS SPEECH RECOGNITION METHOD

摘要

: This invention relates to a speech recognition method, and more particularly to a continuous speech recognition method for recognizing phonemes in continuous speech. In a matching region in which time-series analytical points of a continuous input speech pattern and a standard pattern exist, a two-dimensional mask is moved on the matching region and is set so as to include a plurality of analytical points of each of the input speech pattern and the standard pattern. On the basis of analytical results at the analytical points within the mask, the degree of similarity between the input speech pattern and the standard pattern is calculated in a mask unit. The input speech pattern is thus successively recognized in time series.