摘要 |
<p>Phoneme feature parameters are extracted from input digital speech signals by means of LPC analysis. Phonetic segments (ZE, EE, ER, RO, OO, OQ) having phonetical meanings are obtained together with similarities to prescribed basic phoneti c segments from the feature parameters to be passed through nodes (S0 - S4) of transition networks provided for each word. In passing the nodes, scores (ql - q4) for similarity Sj of predetermined segments (ZE, EE, RO, OO) of the corresponding phonetic segments are made in selective scoring and the accumulation of the scores is used for recognition of continuous word speech.</p> |