is not indicated or on-line with automatic speech recognition system (unsupervised or on-line adaptation of an automatic In order to prevent incorrect adaptation for the recognized word in the speech recognition systems), confidence measure (confidence measures) is used, or one of the recognized phoneme (recognized phoneme), the number of phonemes, a word, the number of words or any utterance (utterance) the speaker (speaker) user response (reaction), and the analysis is used to determine the adaptation of the speaker independent model set for the adapted model set, if the adaptation is performed, these recognized This adaptation is performed as part of a very strong fire or ignite such a recognition. Further, the recognition rate (recognition rate) is (significantly) reduced without, verification of the speaker adaptation performance or increase in order to ensure that maintaining the same level (speaker adaptation performance) is proposed.
展开▼