Stereo Geometry Measurements

Abstract

Methods of evaluating accuracy of geometry of stereophonic sound reproductions involve generating sounds—such as speech or clicks—over a systematic array and then having observers attempt to locate the virtual sources of such sounds on playback over a stereo system. Multiple-row sound source arrays previously have been used. Distance and vertical localizations are poor so lateralization was evaluated with a single row of sound sources. Using the root-mean-square error divided by the array spread as a “criterion of error,” Table I evaluates direct listening (“live”) and the various reproducing systems indicated. TABLE I. Conditions Abbreviation Criterion of error 1. Direct listening Live 0.042 2. Stereo-directional microphone, 2 sound tracks, 3 speakers SD-2-3 0.11 3. 3 Spaced microphones, 3 electrically independent channels 3-3-3 0.123 4. 2 Spaced microphones, 2 sound tracks, 3 speakers 2-2-3 0.137 5. 2 Spaced microphones, 2 sound tracks, 2 speakers 2-2-2 0 30 Playback array subtended nearly 90 deg at the observer. For condition 5, errors increased grossly as the observer moved away from the axis of symmetry, only slightly for the other reproducing systems, and none for the “live” condition.