Abstract:

In everyday listening situations, binaural information enhances the ability to understand speech messages in multitalker environments. For applications to virtual auditory technology in such environments, the importance of preserving characteristics of the natural listening environment is not clear. The purpose of this study was to assess listeners' ability to monitor target (T) messages in the presence of synchronous competing (C) messages in an anechoic and a reverberant environment in three modes: (1) through loudspeakers at (plus or minus)45(degrees) and (plus or minus)90(degrees) with the subject seated in the actual room; (2) through the same loudspeakers with the subject listening remotely to the stimulus presented to a manikin in the actual room; and (3) with the environment simulated to create virtual sources at (plus or minus)45(degrees) and (plus or minus)90(degrees) azimuth under earphones. Fourteen subjects listened to messages selected from four lists of 2034 ten-syllable sentences. In every listening condition, the T message was presented 40 times to the subject in the presence of 0, 1, 2, and 3 C messages; subjects recorded the T messages. Results demonstrate that the ability to understand the T messages decreases as the number of C messages increases. As expected, speech recognition is better in the anechoic environment than the reverberant environment when C messages are present. Performance is degraded whenever the subjects do not listen with their own ears. [Work supported by the U. S. Army Research Lab.] XXSU PP