Test Signals for the Measurement of Non-linear Distortion of Sound Signals.

When non-linearity is present in sound signal circuits, the quality of the reproduced programme may be impaired. It would be convenient if some objective measure of the non-linearity could be used to estimate, with reasonable accuracy, the degree of subjective impairment that the programme would suffer. This would then reduce the need for listening tests in the quality assessment of sound programme circuits. It has long been recognised that the conventional total harmonic distortion measurements are not in good agreement with the results of subjective listening tests. This thesis presents an investigation of new objective methods of using pseudo-random noise as a test signal in the measurement of sound signal non-linearity. The degree of correlation between the results of these non-linearity measurements and subjective estimates of sound programme quality impairment is examined. Subjective listening tests were conducted to assess the sound programme quality impairment introduced by several non-linear circuits, including circuits in which non-linearity was predominant at high signal-frequencies. Possible methods of measuring the non-linearity products of a full audio bandwidth noise-like test signal are discussed and detailed descriptions are given of the development of two novel methods of measurement which use comb-filter techniques to separate the non-linearity products from the test signal. A single comb-filter method, which also requires frequency shifters, is first described and possible ways of improving the method are discussed. An improved measurement method, referred to as the double comb-filter method, is then developed. Both comb-filtering methods gave measurement results which agreed well with those of the subjective listening tests, and this agreement was much better than that obtained by measurements of total harmonic distortion. During the work, emphasis was placed on the importance of developing measurement methods which could be relatively inexpensive to instrument. Possible ways of applying the double comb-filter method to. the measurement of other sound signal distortions (e. g. "wow" and "flutter", quantising distortion and linear distortion) are discussed. Apparatus developed for the double comb-filter method uses mainly digital processing, and with the advent of inexpensive large-scale digital integrated circuits, the cost of instrumenting the method is expected to be less than for the total harmonic distortion method used at present