| 摘要 |
The invention relates to a method for preparing, operating or adapting a voice-driven control device for operating a technical device (52). An audio signal (sa) consisting of at least one word uttered by a speaker is fed to a first signal input (10) of an authentication device (2). A speaker recognition operation is carried out on the basis of an authentication attempt (2) in order to establish a speaker as an individual or as an objectivated group of speakers to which the speaker is to be assigned through objectivated criteria of the audio signal to be traced back to the speaker, and in order to deliver a corresponding output signal (a, a*). A profile (33) corresponding to the established speaker or the objectivated group is selected (2,31) from a number of stored profiles (32,Pi) with the aid of the output signal (a, a*) from the authentication, and the selected profile (33) is integrated or loaded into a recognition environment (4) in order to adapt the recognition environment to the established speaker or the objectivated group. Each of the stored profiles (Pi) and the integrated or loaded profile (33) contain parameters for influencing at least one recognition of a word sequence (42) provided in the recognition environment (4). |
