dorothea_round
Dorothea Wendt

Scientist, PostDoc

mail@eriksholm.com

Pupillometry, measuring the pupil dilation, can be used as a proxy for listening effort. Eriksholm’s work with this method has the objective to evaluate the potential benefits of signal processing applied in current hearing aids.

Recently, different studies have examined the impact of a noise reduction (NR) scheme on effort mobilization in people with hearing impairment. Listening effort was investigated by measuring the participant’s pupil dilation while doing a speech recognition task. Pupillometry is a commonly used method for assessing the effort involved in performing a task, such as speech recognition in background noise.


listenplusfig2_ramme

Figure 1: Example of normalized pupil curve averaged across a group of listeners. The pupil size was measured while people listening to sentences in noise. The PPD is calculated as the maximum pupil dilation during stimulus presentation. Figure from Wendt et al., (2017)

Measuring Peak Pupil Dilation

The Peak Pupil Dilation (PPD) is defined as the maximum pupil dilation during the time interval between sentence onset and the noise offset (see Figure 1). It is a commonly used indicator of listening effort in such an experimental paradigm. The larger the PPD, the higher the listening effort.

In adverse and noisy listening situations, listening effort is expected to be high, which would be reflected in an enlarged pupil dilation. By applying a NR scheme (including directional microphone use and NR), the hypothesis was that listening effort can be reduced for people with hearing impairment, as indicated by a significant decrease in pupil dilation.

A second study was performed that also focused on the impact of the signal processing on effort over a broader range of listening situations (i.e. broad range of SNRs). This study demonstrates that listening effort changes in a non-monotonic way as a function of the SNR with maximum effort occurring at approximately 50 percent intelligibility. Furthermore, reduced effort was observed again at ecological and high SNRs reflecting everyday communication (with speech intelligibility above 80 percent). Interestingly, in more adverse listening situations where speech intelligibility was reduced, listeners even mobilized more effort with the active NR scheme. This study was part of a PhD project within the LISTEN project.

listenplusfig

Figure 2: PPD (in mm) averaged across a group of people with hearing impairment. NoNR = noise reduction was turned off; NR = NR scheme was turned on. L50 = SNR corresponding to 50% intelligibility; L95 = SNR correspond to 95% speech intelligibility. Figure from Wendt et al., (2017).

Results

Figure 2 shows reduced PPD, and thus a reduced effort mobilization, at both 50 percent and 95 percent speech intelligibility with an active NR scheme. Interestingly, listening effort was reduced in a condition where speech intelligibility was at ceiling. In other words, listening effort was reduced in a situation where people were able to recognize most of the speech even without the hearing aid signal processing. This situation, which was tested at a signal-to-noise-ratio (SNR) that reflects everyday communication, is often referred to as ecological listening conditions. For more detailed information about the study and the results see Wendt et al. 2017.

A second study was performed that also focused on the impact of the signal processing on effort over a broader range of listening situations (i.e. broad range of SNRs). This study demonstrates that listening effort changes in a non-monotonic way as a function of the SNR with maximum effort occurring at approximately 50 percent intelligibility. Furthermore, reduced effort was observed again at ecological and high SNRs reflecting everyday communication (with speech intelligibility above 80 percent). Interestingly, in more adverse listening situations where speech intelligibility was reduced, listeners even mobilized more effort with the active NR scheme. This study was part of a PhD project within the LISTEN project.