Effects of masker type and reverberation on speech-in-noise recognition thresholds and listening effort as indexed by pupil dilation responses

Pupil dilation responses have emerged a promising way to index listening effort required to understand speech in noisy environments. In this study, pupil dilation responses were captured with a portable eye tracking glasses while measuring speech recognition thresholds in noise and reverberation with the matrix sentence test. Three different masker types included stationary speech shaped noise, phase doubled babble and babble talk composed of speech from two other languages than the target talker. Measurements were made in a standard listening room with dry stimuli without reverberation as well as by using auralization of a real room with 1.5 second reverberation time. Twenty listeners participated to the study. The participants included Finnish, German and both native and non-native English speaking people. Overall, speech recognition thresholds were higher in reverberation compared to the dry condition and this result was also well reflected by the peak dilation responses, that showed on average greater and delayed dilations in reverberation than in the dry conditions. However, the patterns of speech recognition thresholds (SRTs) and pupil dilation responses between the masker types were somewhat different. Speech recognition thresholds in both babble noise and babble talk were higher than in the speech shaped noise, but peak dilations were higher only in babble talk, whereas similar responses were observed for both speech shaped and phase doubled noise. Therefore, pupil dilation responses may provide another viewpoint to the speech-in-noise recognition performance.