The performance of spectral quality measures

Two different classes of quality measures are discussed and compared: absolute and relative measures. The relative class to which the prediction error belongs has many different approximations and equivalents, like the spectral distortion and the likelihood ratio. This measure is based on time series theory and can be written as a relative error in the frequency domain. It is useful in many applications. It will be compared to some absolute measures. To that class belong a squared difference measure on the integrated spectrum, that gives equal weights to all frequencies, and also a measure that is based on the squared difference between impulse responses. The absolute class has only a few practical applications, mainly in speech.