Abstract:

The binaural system helps listeners to understand speech at low signal-to-noise ratios, presumably via time-varying binaural cues. However, the binaural system is often described as sluggish, because listeners cannot follow rapidly changing binaural cues. The speech in such experiments may have been articulated slowly enough for the binaural system to temporally resolve critical formant transitions. Alternatively, changing signal spectra may be analyzed by a nonsluggish aspect of the binaural system. In the former case, binaural unmasking of speech will depend upon speech rate. In a preliminary study, listeners' discrimination of changing signals in noise was measured as a function of the rate of frequency change. Masked discrimination thresholds were taken for NoSo and NoS(pi) using a 2I-2AFC task. The signals consisted of ascending or descending three-tone glissandos (400, 500, and 625 Hz), presented repeatedly and continuously during a 1.6-s broadband white noise burst. Rate of frequency change was controlled by altering the duration of each individual tone (32 to 100 ms). The difference between NoSo and NoS(pi) was smaller for rapidly changing signals. Thus the binaural unmasking of speech should be dependent upon speech rate. However, the rate of articulation sufficient to abolish masking release is not known.