| 摘要 |
A speech-controlled device produces a visual representation (via a typewriter, printer or display device) of a spoken input. The device operates in real time without speaker-dependence, and comprises a fast speech analyzer (FSA) for analyzing the speech input as relatively fast speech and a slow speech analyzer (SSA) for analyzing the speech input as relatively slow speech. In a first embodiment, a selector is responsive to a determination that the speech input is relatively fast or slow speech for selecting the output of the FSA or SSA, as appropriate. In a second embodiment, an enable circuit is responsive to a determination that the speech input is relatively fast or slow for enabling a corresponding one of the analyzers for operation. The FSA comprises a first section which derives word features, a second section which derives word shapes, and a comparator for comparing word features and word shapes so that a word identification is made when coincidence occurs between the word features and the word shapes. |
