9th Speech in Noise Workshop, 5-6 January 2017, Oldenburg

Does listening effort modulate speech envelope entrainment?

Lien Decruy(a), Jonas Vanthornhout, Tom Francart
KU Leuven, Leuven, Belgium

(a) Presenting
(b) Attending

Studies have shown that cortical neural activity is entrained to the amplitude envelope of running speech. Given that the envelope is a primary cue for speech intelligibility, the degree of envelope entrainment may reflect different levels of speech intelligibility. Consequently, envelope entrainment in function of signal-to-noise ratio (SNR) may provide an objective correlate of speech intelligibility which can be used for the audiological assessment of populations such as young children. In our study, we measured envelope entrainment in function of SNR by calculating the correlation between the actual and reconstructed envelope from EEG-signals.

Interestingly, several subjects showed a decrease in envelope entrainment at SNRs where they achieved a high speech intelligibility score. This decrease contradicts our hypothesis since we assumed maximal entrainment when speech is completely intelligible. On the other hand, studies have found that high-level processes can modulate envelope entrainment. Taken this into account, listening effort may explain the decrease in entrainment on high SNRs. In challenging situations, a person will invest extra effort to restore the degraded speech signal, while having a conversation in quiet requires almost no effort. The most widely used method to measure listening effort is the dual task paradigm. This method relies on the theory that cognitive resources are limited and suggests that performance on a secondary task will deteriorate when the primary task is very demanding. Despite this theory, dual-tasks often involve two different sensory modalities which results in the allocation of different resources instead of shared. Furthermore, several studies use the target of the primary task, for example a word, as input for the secondary task. In this case, the performance on the latter task will not only be influenced by the effort needed in the primary task but also by the extent in which the word was heard. Therefore, we developed a new dual task.

Thirteen normal hearing and 2 hearing impaired, young subjects participated in our study. The primary task of our dual task involved a speech-in-noise test where the Flemish Matrix sentences were presented at fixed speech intelligibility levels. Our secondary task was a verbal-memory test in which response time was measured. To compare with our objective measure, we presented the same sentences on the same levels during an EEG-experiment. Additionally, we also investigated the reproducibility of our objective measure.

Last modified 2017-01-04 23:51:47