| 摘要 |
<p>The method consists in analysing phonemes of the phonetic signal at regular intervals of time, comparing the phonemes with a series of reference phonemes, selecting the reference phoneme closest to the phoneme of the phonetic signal and generating a visible indication of the selected phoneme in real time on a monitor (20) to allow the speaker to adapt his speech to the response of the apparatus.</p><p>The apparatus comprises a series of analysers (A<Sub>i</Sub>, A<Sub>2</Sub> ....An) for a phonetic signal segment for generating a series of vectors (V) to be compared by a set (13) of comparators (C<Sub>1</Sub>, C<Sub>2</Sub>.....C<Sub>m</Sub>) with a plurality of reference vectors corresponding to the various reference phonemes.</p><p>A control unit (19) receiving the result of the comparison controls the visual display on the monitor (20) of a histogram indicating the selected phoneme and also a merit value (P) and a stability value (ST), of the recognition.</p><p>A lexical recognition unit (22) is adapted to link individual phonemes together to recognise words and to display the words on the monitor (20).</p> |
