Investigating Calculated Loudness ISO 532 for Evaluating Impulsive Sounds

Because of the multitude of impulsive sounds that can occur in automobiles it would be valuable to have a method for evaluating these sounds that relates well to customer response. Loudness Calculation based on ISO 532 Method B ( Zwicker' s method) was investigated for its potential as such a method. Problems occurred in obtaining valid 1/3 octave band spectra from conventional real time analyzers for the input to the calculation routine. A conflict arises between the short averaging times needed to track the impulse and the uncertainty principle (BT product rule) for the lower frequency bands. Since Zwicker' s method is based on critical bands, a four-pass analysis procedure was devised using an external filter set for the width of a critical band. The calculation routine was modified to directly accept critical band data. Example plots of Calculated Loudness versus time are presented showing the affects of selecting long averaging times, of not obeying the uncertainty principle, and of using the four-pass analysis method with an averaging time of 1/64th of a second. It is concluded that to properly use Zwicker' s Calculated Loudness method for impulsive sounds will require an analyzer based on critical bands rather than the standard 1/3 octave bands.