作者: Hiroaki Hattori
DOI:
关键词:
摘要: A speech recognition system, in which partial reference patterns, and cumulative similarities of these are stored a temporary pattern memory. The patterns to be used as subjects similarity computation with an input that has its feature quantities extracted by analyzing unit. counting unit counts having corresponding higher than threshold value computing computes pruning from correspondence relation between the number have exceed threshold, threshold. similarity, respect quantities, greater pruning. determining determines, result recognition, word corresponds one highest similarity.