The Weighting Game

Many systems have been developed for weighting noise data to provide meaningful indicators of human responses. Beginning with the weighting networks of the standard sound-level meter, numerous more sophisticated weighting methods have been devised to improve correlation with subjective reactions. None has proved too successful. Some of them even disagree significantly as, for example, the Stevens and Zwicker loudness calculations. Also, the proposed N weighting does not indicate PNdB accurately as intended. The calculations of these weighted levels are all too complicated for general use, but each correlates with sound levels extremely well. The reason for the high correlation becomes clear when it is realized that the octave-band SPL's of a noise are predictable from its sound levels with surprising accuracy. This fact even makes possible the expression of sound transmission loss, permissible audiometric room levels, etc., in terms of sound levels. Since searching for better weighting methods is delaying solutions of current problems and since, for practical purposes, sound levels are just as accurate, sound levels should become the preferred weighted levels for use in dealing with current problems, including the puzzles of human responses.