Energetic and informational masking in a simulated restaurant environment.

Participants were seated at a central table for two in a virtual restaurant, simulated over headphones. They listened to the person across the table. Speech reception thresholds (SRTs) were measured as a function of the number of interfering sources distributed across other tables in the room; those sources were either speech-shaped noises or competing speech. The restaurant either was enclosed by acoustically reflective surfaces or was anechoic. Variations in SRT for speech-shaped noises were accurately predicted (r = 0.96) by a model of spatial release from masking based on the additive combination of better-ear listening and binaural unmasking. However, SRTs for interfering voices followed a different pattern. A single interfering voice gave a lower SRT than a single speech-shaped noise source (by 6.3 dB in anechoic conditions and 1.2 dB in reverberant conditions). This difference can be attributed to the effects of dip listening and to the exploitation of differences between voices in fundamental frequency (F0). SRTs for two interfering voices were markedly higher than for a single voice, particularly when the interfering voice was the same as the target voice. Multiple speech interferers produced more masking than multiple noise interferers. This effect can be attributed to informational masking (IM). These results indicate that current models require some elaboration before they will produce accurate predictions of intelligibility in noisy social environments.