摘要 |
PURPOSE:To provide the speech recognizing device which can accurately recognize all of plural different speech patterns. CONSTITUTION:This speech recognizing device recognizes prescribed speech data. This device includes a speech recognizing process means 20 including plural neural network parts 200-1, 200-2... for speech recognition which in advance learn speech patterns differing in features so as to recognize the speech data 100, performs speech recognizing operation for determining whether or not the inputted speech data coincide with speech data to be recognized, and outputs data for deciding adaption of speech recognition, a selecting means 30 which selects the neural network part for speech recognition having the highest adaption of speech recognition according to adaption decision data outputted from the respective neural network parts 200-1, 200-2..., and an output control means 40 which outputs the speech recognition result from the neural network part for speech recognition selected by the selecting means 30. |