Abstract
This study explores the factors that influence the masking of narrow-band signals in reverberant sound fields. Rectangular rooms were simulated using the image method with a rigid sphere modeling the head. The signals were third-octave noise bands presented in a continuous broadband noise masker. For a given room simulation, the signals and masker were convolved with impulse responses obtained in the room at multiple azimuth angles and distances, for subsequent presentation via headphones. Adaptive forced-choice procedures were used with the resulting signals to find monaural and binaural thresholds in listeners with normal hearing. The benefit of separating the masker and signal spatially in the simulation was reduced by reverberation, as expected, although the effects were complex. Among the predictions confirmed by these experiments is that reverberation actually improves thresholds when the masker is closer to the listener than the signal, apparently due to both an increase in monaural signal power a decrease in interaural signal correlation. With masked thresholds obtained over the frequency range of the articulation index, the results of these experiments will allow quantitative predictions of speech intelligibility as a function of room charcteristics and speech and noise source locations. [Work supported by NIH.]
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have