摘要 |
<p>The method involves determining vocabulary entries in a speech signal from a preset vocabulary, and determining the speech signal from the vocabulary entries. A sampling of determined speech signal information is reduced, and a state sequence of entries of the preset vocabulary is reduced. Sampling values are determined by sampling from a time signal of the speech signal. Characteristic vectors are determined by characteristic extraction from the sampling values. A sampling rate is reduced for reducing number of sampling values.</p> |