Perceptual Weighting of Masking Release due to Voice-Gender Difference and Spatial Separation in a ‘Cocktail Party’ Environment
Yonghee Oh, PhD
Problem Statement: Voice-gender difference and spatial separation between talkers are important cues for speech segregation in multi-talker listening environments. The purpose of this study was to systematically investigate the interactions between voice-gender difference and spatial separation cues within cocktail party listening environments in order to better understand their influence on the task of speech segregation.
Methods: Twenty normal hearing (NH) listeners were tested on speech recognition threshold measurements in competing speech at various listening conditions with multiple talkers, of either the same or mixed genders, that were presented from multiple spatial locations (0°, ±15°, ±30°, ±45°, and ±60°). Masking release benefits by either a voice-gender difference cue (voice gender release from masking, VGRM) or a spatial separation cue (spatial release from masking, SRM) were calculated.
Results: Average results revealed an inverse relationship between VGRM and SRM. At the spatial separations less than ±30°, VGRM was larger than SRM, and SRM grew larger with increasing separations while VGRM decreased. VGRM and SRM intersected and became equal at only one spatial configuration just slightly larger than ±30°. At the separations larger than ±30°, the opposite effect was recorded and SRM was larger than VGRM.
Conclusions: The findings in this study suggest that there is a clear equal perceptual weighting of VGRM and SRM at only one listening condition and an unequal perceptual weighting at all other listening conditions tested. The equal perceptual weighting between VGRM and SRM occurred at their intersection, suggesting that this listening condition may be optimal to improve a normal hearing listener’s speech segregation performance in cocktail party listening environments.