A Statistical Basis for Objective Measurement of Speech Levels

1.1 Object of Study The goal of this study is determining a speech level measure ideally having the following properties: (1.) It is objective, and is not based on the judgment of an observer. (2.) It is based on measurements made only while speech is present and is not influenced by long silent intervals. (3.) It is expressible as a single number. (4.) It varies on a db-for-db basis with attenuation or amplification of the voice signal. (5.) It is not a function of an arbitrary convention used to take the measurement, such as the value of a threshold in a meter. 1453 1454 T H E B E L L SYSTEM T E C H N I C A L J O U R N A L , S E P T E M B E R 1965 (6.) It is not influenced by singular loud transients on the voice circuit. (7.) It is easily and reliably obtained. The specification of the level of a signal implies a description of certain physical properties of the amplitude of the waveform. The loudness of a signal is a measure of the volume of a sound as perceived by a listener. Although it may be possible to correlate level measurements with loudness, no attempt will be made to do so in this study. 1.2 Outline of Report In seeking a measurement satisfying the above requirements, an analysis is made of the statistics of speech levels as they appear above a threshold. This analysis, appearing in Section II, shows the logarithms of these levels are nearly uniformly distributed. Section I I I indicates that the peak amplitude occurring in a speech sample satisfies most of the requirements listed above.