Re-Vision of the Speech-Privacy Calculation

Abstract

The articulation index used in telephone communication, the listening equation for SONAR, and the acoustical privacy calculation of architectural acoustics are all founded on a computation of an excess of signal level over noise level just sufficient to permit some stated detectability. From this viewpoint, data compiled for “Speech Privacy in Buildings” [Cavenaugh, Farrell, Hirtle, and Watters, J. Acoust. Soc. Am. 34, 475–492 (1962)] have been reviewed for possible simplications in their procedure for estimating acoustical privacy. In lieu of the “dot counting” method, ratings of sound isolation here investigated include the sound transmission class (STC), a similar index being discussed for international standardization (IISO), and a frequency-weighted sound isolation; as measures of ambient noise, in lieu of curves to be fitted for different spectrum shapes, sound level (A) and frequency-weighted octave-band sound-pressure levels have been tested. The reported judgments of be correlated at least as well with ratings based on these presently considered measures of sound isolation and noise as with those originally employed. Accordingly, ways of substituting one kind of rating for another have been derived.