摘要 |
The invention relates to a method for estimating speech quality, wherein a reference speech signal (101) with a reference speech spectrum (X) enters a telecommunication network resulting in a test speech signal (102) with a test speech spectrum (Y), the method comprising the steps of identifying a number of perceptually dominant frequency sub-bands (bi) in one of the reference speech spectrum (X) and the test speech spectrum (Y), computing an intensity scaling factor (c i ) for each identified sub-band (bi) by minimizing a measure of the intensity difference between those parts of the reference speech spectrum (X) and of the test speech spectrum (Y) that correspond to the respective sub-band (b i ), multiplying the test speech spectrum (Y) with each intensity scaling factor (c i ) thereby generating a number of scaled test speech spectra (Y i ), selecting one scaled test speech spectrum, and computing the difference between the selected scaled test speech spectrum (Y sel ) and the reference speech spectrum (X), the difference being indicative of the speech quality of the test speech signal (102). The invention furthermore relates to a method for estimating speech quality in which the reference and the test speech signal are aligned and to a method for estimating speech quality that deals with interruptions and speech loss. |