15th Speech in Noise Workshop, 11-12 January 2024, Potsdam, Germany 15th Speech in Noise Workshop, 11-12 January 2024, Potsdam, Germany

P57Session 1 (Thursday 11 January 2024, 15:35-18:00)
The impact of across-frequency coherence on speech intelligibility in young and older normal-hearing listeners

Helia Relaño-Iborra, Torsten Dau
Hearing Systems Section, Department of Health Technology, Technical University of Denmark, Lyngby, Denmark

People with clinically normal hearing thresholds may still encounter challenges understanding speech in adverse conditions, yet identifying such suprathreshold deficits has proven challenging. Motivated by auditory-modeling studies proposing an analysis of across-frequency coherence in the backend of speech-intelligibility (SI) prediction frameworks, we explored a ‘distortion-sensitivity’ approach to characterize listeners’ perception of across-frequency phase coherence as a potential suprathreshold factor influencing speech perception. While a loss of phase coherence across frequency bands, such as through phase jittering, has been shown to be detrimental for SI in young normal-hearing listeners, less is understood about the sensitivity of older individuals and those with listening challenges to such distortions.

We manipulated across-frequency coherence by independently adding a random phase component to each frequency channel of a speech-and-noise mixture presented at a signal-to-noise ratio of 5 dB. SI was measured based on the spread of the distribution of phase values. Additionally, we obtained the just-noticeable-differences (JNDs) for across-frequency phase jitter, as well as speech reception thresholds (SRTs) for undistorted speech in the presence of stationary, fluctuating and speech-like maskers. Data from 10 young and 10 older NH listeners revealed lower JNDs in the older listener cohort, suggesting a reduced sensitivity to across-frequency coherence. However, no changes in the perception of jittered speech was observed. The results from this experimental work will be translated into model parameters, enabling the evaluation of a suprathreshold distortion component within a computational framework for the predicting the measured SRTs.

Last modified 2024-01-16 10:49:05