作者: Hsaio-Wuen Hon , Kuansan Wang
DOI: 10.1121/1.1697777
关键词:
摘要: A speech recognition method and system utilize an acoustic model that is capable of providing probabilities for both a large unit sub-unit. Each these describes the likelihood set feature vectors from series representing signal. The formed plurality sub-units. At least one sub-unit probability at on are used by decoder to generate score sequence hypothesized words. When combined, sub-units associated with all determine span fewer than in vectors. An overlapping decoding technique also provided.