| 摘要 |
In the method for the segmentation of speech an acoustic speech signal is converted into N signals Si(f), each signal pertaining to a time interval i of N successive time intervals (1</=i</=N). For successive time intervals i a function cij is then derived which is a measure for the agreement between the signals in the time intervals i and j. The middle mi of the pertaining function cij is determined for each time interval and those values i are determined which at leastapproximately correspond to zero-axis crossings with the same sign in a function di, di indicating the difference between mi and i. The signal segmented in this way can be used for the derivation of diphones, so that a library of diphones can be built up, or it can be used for the recognition of the speech utterance. Arrangements which realize these functions are described. |
