In the measurement based approach it is desirable to compare a natural signal with a synthetic signal to see how accurately it is possible to simulate the natural one. In auditory speech, for example, we might compare spectrograms to see how closely the patterns of formants (resonances) agree in frequency and temporal characteristics. Regarding suprasegmental information, we might compare the pitch contours of natural and synthetic sentences. For the auditory channel, many standard techniques for analysis of these characteristics are available. For example, FFT, cepstrum, and linear prediction (LPC) can be used to extract formant information. For the visual channel, the situation is less advanced and much less standardized.