11th Speech in Noise Workshop, 10-11 January 2019, Ghent, BE

Switching attention and integration of binaural information: Examining the effects of masker types and binaural listening on perception of interrupted speech in noise

Shiran Koifman(a), Stuart Rosen
University College London, UK

(a) Presenting

Over the last few years we have been investigating the more general utility of a task which we have shown to be highly sensitive to the effect of aging for speech maskers when compared with a standard speech in noise task. In this task, the target speech is interrupted at a fixed modulation rate (5 Hz), with successive segments of the target being switched from ear to ear. An adaptive procedure is implemented by varying the duty-cycle (DC), which is the proportion of time the signal is ‘on’ in each modulation period, in order to find the proportion of speech required to understand 50% of the keywords (i.e., the speech reception duty-cycle threshold, SRdT). A masker is interrupted in the same way, and alternated between the two ears out-of-phase with the target speech, resulting in alternated segments of both target and masker signals between the two ears, with only one stimulus present in each ear at any given time. This task appears to exploit some higher-level cognitive aspects of listening not probed by simpler tasks, and is believed to necessitate the listeners’ ability to switch attention and integrate short-term auditory information between the two ears.

Here, we aim to examine several parameters that might affect performance in the task, namely: the effect of masker types (speech vs. non-speech), the extent to which listeners are actually obtaining information from both ears as opposed to attending to one ear only, and also the influence of speech material.

To examine the effect of masker type, listeners were presented with simple sentences in three types of maskers; unrelated connected speech, and two non-speech maskers which were extracted from the original speech maskers and varied in their amount of “speech-like” characteristics (from high to low): (1) a single band vocoded speech with natural mix of periodicity and aperiodicity [1]; (2) amplitude modulated speech-shaped-noise. To examine the ability to make use of alternated stimuli, we compared the listeners’ performance in two listening configurations: (1) binaural in which the stimuli are fully preserved when segments of the stimuli from both ears are combined, (2) and a monaural configuration where only the information in one ear is presented. Lastly, the influence of speech material was explored by comparing performance with CRM-like sentences.

A fuller understanding of the abilities exploited by this task may make it useful in helping to disentangle the reasons why various groups of people experience difficulty in listening in noisy situations.

This work was supported by Action on Hearing Loss, UK.

[1] Steinmetzger, K., Rosen, S. (2015). The role of periodicity in perceiving speech in quiet and in background noise. J. Acoust. Soc. Am., 138(6), p.3586–3599.

Last modified 2018-12-08 00:23:30