Spectral integration of synchronous and asynchronous cues to consonant identification

Abstract

When speech is presented at poor speech-to-masker ratios, the listener may momentarily receive some spectral regions of the signal at more favorable ratios than other regions. These optimal spectral regions may vary dynamically as a function of time. It is therefore of interest to determine how well listeners utilize cues from different spectral regions that occur asynchronously as opposed to synchronously across frequency. Here, we examined masked identification of consonants under various conditions of 10-Hz square-wave modulation of the speech material. Speech from 100–10<th>000 Hz was divided into either 2, 4, 8, or 16 contiguous bands. In one set of conditions, bands were amplitude modulated coherently, while in the other, even and odd bands were modulated 180 degrees out of phase. These conditions, along with further conditions employing only modulated even or odd bands, allowed performance to be compared between synchronous and asynchronous presentation. This approach is similar to a published checkerboard masking study in which the noise rather than the speech was modulated [Howard-Jones and Rosen, J. Acoust. Soc. Am. 93, 2915–2922]. Preliminary data suggest that many listeners are able to utilize asynchronous speech cues efficiently in all of the band number conditions examined. [Work supported by NIDCD (5R01DC000418).]