| 摘要 |
PURPOSE:To automatically recognize a syllable without uttering a voice by learning a brain wave topography pattern at the time when the utterance of a voice is imaged, recognizing the imaged syllable after learning and showing it automatically. CONSTITUTION:A person who executes learning in a recognition mode is allowed to generate continuously N times one of syllables which a nerve network 4 learns already. When it is being generated, a signal (a) detected by an electrode 1 is detected as a time series brain wave signal (b) by a brain wave meter 2, and based on a trigger signal at the time of utterance by a voice detecting device 8 as a reference, signal brain waves before and after the utterance are added, and it is repeated N times. Its time series preparation potential is converted into a bran wave topography (BT) and processed by a processor 9 and inputted to a nerve network 4 as a topography signal C2, the BT is recognized as a numerical matrix, and the corresponding syllable is transmitted as an output (e) to a display part 6. Subsequently, when the generation of its syllable is imaged, in the same way, the syllable is recognized automatically. |
