Artikel
Amplitude compression on ignored talker increases neural separation of competing speech
Suche in Medline nach
Autoren
Veröffentlicht: | 12. September 2022 |
---|
Gliederung
Text
Cochlear hearing loss is associated with a greater than normal rate of growth of perceived loudness with increasing sound level. Modern hearing aids take on the problem by making low level sounds audible and keeping high intensity sounds comfortable by using amplitude compression. However, amplitude compression has undesired side effects, such as reduction of amplitude modulation depth or abrupt changes in the envelope magnitude at the onsets or offsets of sound. The behavioural and neural consequences of amplitude compression and how they interact with spatial selective attention in a multi-talker situation are at present unclear. Here, we tested 24 normal-hearing listeners in a psychophysically augmented continuous speech paradigm. Listeners attend to one talker (target) while simultaneously ignore another talker (distractor). We embedded short repeats of the used speech in both streams. Listeners had to detect repeats in the target stream while they had to ignore repeats in the distractor stream. We added amplitude compression (compression ratio 1:8; loudness-matched) as signal manipulation to both streams and contrasted (fully balanced) uncompressed versus compressed speech in our paradigm. Neural responses were recorded in the electroencephalogram to obtain temporal response functions (TRF) and predictive accuracies from TRF-based encoding models (neural speech tracking). At the neural level, we asked how compression affects neural tracking and interacts with attention. We found larger neural tracking of the target compared to distractor with a similar effect size as reported in previous literature. Neural tracking of compressed speech was smaller compared to the uncompressed speech. Interestingly, the compression effect was larger than the attention effect. Importantly, the neural tracking index (separation of attended vs. ignored) was larger when ignored speech was compressed versus uncompressed ignored speech. At the behavioural level, participants had expectedly higher accuracy (proportion of correctly detected repetitions in the target stream) for uncompressed versus compressed speech. However, we found no conclusive behavioural evidence for a larger behavioural separation when ignored speech was compressed. Findings suggest that compression impairs the neural and behavioural response to speech. However, a larger attentional separation of competing speech at the neural level when the ignored stream is compressed might prove useful in supporting hearing impaired individuals in their ability to ignore distractions in multi-talker hearing situations.