Background: The amplitude envelope of speech is crucial for comprehension, and cortical activity in the theta-delta bands has been shown to track this envelope. This tracking is thought to reflect responses to transient events in the speech signal, such as sound onsets (rapid increases in sound amplitude) and offsets (rapid decreases in sound amplitude). While onset tracking has been widely studied, offset tracking remains underexplored.

Methods: In this study, three independent datasets from different laboratories were analysed, consisting of continuous EEG recordings during audiobook listening in quiet from British (n=18) [1] and Danish (n=22) [2] participants or in noise from British (n=28) participants. An onset/offset model based on thalamic responses to sound transients in mice [3] was used to extract separately the onsets and offsets present in the broadband envelope of the continuous speech. Linear forward models using either speech onsets, offsets or both onset and offsets as regressors were used to assess separately how well each of the transient events are tracked during naturalistic speech listening. Additional analyses were performed on onsets and offsets extracted from a 16-channel gammatone spectrogram of the envelope to explore if onset and offset processing during speech listening relies on distinct spectral regions.

Results: Model performance, measured by the correlation between predicted and recorded EEG, was significantly above chance for all representations (p<.001), confirming that both onsets and offsets contribute to speech tracking. On the broadband envelope, Wilcoxon tests showed that onset-based models outperformed offset-only models (p<0.001, d=0.69), suggesting that onsets have a stronger effect. However, the combined model performed best overall, surpassing both offset-only (p<0.001, d=0.85) and onset-only (p<0.001, d=0.56) models. This finding highlights the complementary role of offsets in cortical speech tracking. Analyses of the envelope spectrogram are still being finalised, but preliminary results suggest that high-frequency channels may be particularly important for offset tracking.

Conclusion: Overall, our results suggest that sound onsets and offsets play a distinct role in cortical speech tracking both in quiet and in noise, with onsets exerting a stronger influence, but offsets providing significant complementary information.

References:

• [1] Etard O & Reichenbach T (2022). doi:10.5281/zenodo.7086208
• [2] Simon A, Bech S, Loquet G & Østergaard J (2022). doi:10.5281/zenodo.7500806
• [3] Anderson LA & Linden JF (2016). J Neurosci 36(6), 1977-1995.