Objective Measures of Peak Clipping and Threshold Crossings in Continuous Speech

This study reports data on the statistics of instantaneous speech levels in continuous speech samples, with special emphasis on threshold crossings and other quantities related to peak clipping. Since a standard measure of specifying the amount of peak clipping is lacking, the study begins by arbitrarily defining the clipping level in terms of an objective speech level measure, the equivalent peak level (epl). 1 Then, measures of time spent above various thresholds (or clipping levels) are reported. Finally, a method is suggested for interpreting clipping as speechcorrelated noise. Empirical measures are given for the amount of noise introduced as a function of the amount of clipping, and signal power lost because of clipping. This paper does not address itself to subjective measurements related to clipping of the speech signal. The subjective effects of clipping must 933 934 THE BELL SYSTEM TECHNICAL JOURNAL, APRIL 1972 eventually be considered in a treatment of the broad problem of determining clipping performance objectives. It is also necessary to examine the objective effects of clipping on the transmitted signal so that guidelines can be established for design and operation of transmission systems. This paper is directed toward providing data to assist in engineering considerations of speech circuits.