| 摘要 |
A speech recognition system recognizes voices from sequence data which is expected to be input in a regular, predetermined order. The speech recognition system includes a unit generating a recognition reference word by adding items of dummy data to each item of the sequence data, and a unit recognizing and outputting from among a set of recognition reference words that recognition reference word which is the most similar to an externally input voice. The recognition reference word is output as a recognized result on the basis of a feature of the input voice. Moreover, the dummy data are neighboring data which lie in a neighborhood of all items of the sequence data, thus forming the set of recognition reference words available for speech recognition.
|
