Method for speech recognition, particularly of names, and speech recognizer

摘要

A voice recognition method, especially for names, has an input phase for establishing a voice recognition-vocabulary and includes the steps (a) of text-phoneme conversion of a word inputted as text, in particular names, for gaining at least one phoneme sequence provisionally assigned to the word in the voice recognition vocabulary, followed by (b) phoneme-voice-conversion and acoustic output of the obtained phoneme sequence. A verification input of the user is then (c) evaluated by assuming the provisional phoneme sequence is a valid phoneme sequence in the voice recognition vocabulary and then either returning to step (a) for getting another provisional phoneme sequence or to step (b) for phoneme-voice-conversion of an alternatively prepared other provisional phoneme sequence in response to a non-confirmed verification input. An Independent claim is given for a voice-recognition device for speaker-independent voice recognition.