The standardization of monaural phase

Abstract

This paper redefines and examines the problem of monaural phase distortion in audio systems. Both technical and psychoacoustical aspects of the problem are considered. The redefinition of phase distortion is based upon adopting the acoustical signal at the input of the system as the standard reference. The sources of phase distortion in audio systems are examined with reference to this standard. A review of early and recent psychoacoustic experiments concludes that monaural phase effects can significantly affect the quality of a perceived sound. A signal detection model is used to determine a conservative equivalency between frequency dependent phase distortion and amplitude distortion. It is suggested that a system which maintains a 3 dB tolerance in frequency response should also maintain a 17° tolerance in phase shift. A recent experiment is shown to confirm the conclusion drawn from this model. Absolute phase (polarity error) is evaluated with reference to amplitude distortion in light of recent experiments. It is concluded that at high sound pressure levels absolute phase error is more detectable than 11.5 percent intermodulation distortion. A set of standards for various types of audio equipment is defined, which if implemented would eliminate absolute phase error in audio systems. Finally, the importance of these findings with reference to realism of reproduced sound is commented upon.