Towards objective markers of speech understanding based on neural coding and perception of spectrotemporal sound features
Background: In the audiological practice, much variability in speech understanding is observed between patients, despite similar (normal or increased) hearing thresholds. Speech perception could be influenced by three different individual factors (audibility, cognition and suprathreshold coding). A factor that could influence suprathreshold coding would be cochlear synaptopathy (CS; damage to the synapses between the inner hair cells and the afferent nerve fibers), which could be measured with the envelope following response (EFR). This objective EEG-test would measure the ability to process the temporal envelope (ENV) cues in speech signals. Nevertheless, speech signals are not only temporally modulated, but also spectrally, where the temporal fine structure (TFS) plays an important role. Research with behavioural psychoacoustic measurements has shown that spectral (SM) and spectrotemporal modulated (STM) stimuli appear to be good representatives of realistic speech signals (and TFS processing).
The study aimed to analyze: (1) Whether behavioural psychoacoustic measurements with S(T)M-stimuli are related to TFS-processing and speech perception. (2) Whether an objective frequency following response (FFR) with SM stimuli could offer information about TFS-processing and speech perception.
Methods: We designed a battery of psychoacoustical detection thresholds and electrophysiological FFR measurements based on spectrally modulated stimuli carrying well-defined TFS cues. The protocol was conducted in three groups of participants: younger normal-hearing individuals (yNH group, n=15), older individuals with normal hearing (oNH, n=16) or mild-to-moderate hearing loss (oHI, n=14). Speech in noise (SPIN) scores were measured with Matrix tests presented under several frequency-filtering conditions.
Results and conclusions: We will present and discuss all the results obtained from this large data-set. One important result was that the strength of FFRs was strongly reduced in older compared to younger listeners, and negatively correlated to age. In contrast, we found no differences in FFR strength between older listeners with or without sensorineural hearing-loss. These data suggest that cochlear damage per se does not impact the neural coding of TFS cues, but that CS could be an important factor for TFS-coding capacities. However, we observed no relationships between these neural measurements and SPIN, nor between psychoacoustical detection thresholds of spectral envelope shapes and SPIN in either group.
These joint psychophysical and electrophysiological results call for further work to design a psychoacoustical test that would provide a robust behavioral proxy of TFS-coding fidelity for such spectrally-shaped sounds. Overall, this project paves the way toward novel non-invasive biological markers of CS based on the neural encoding of spectral cues of sounds.
Funding: Work supported by Fondation pour l’Audition (FPA 2020-005F2) and European Research Council (ERC-StG-678120, RobSpear).