Hearing is essential for communication, but as people age, understanding speech in noisy environments becomes challenging. Age-related hearing loss contributes to this, but the alteration of brain mechanisms involved in the analysis of complex auditory scenes - such as attention, memory, or inhibition of irrelevant information - also plays an important role. Speech perception can be enhanced when the listener can both hear and see their interlocutor. Various types of visual signals can promote audio-visual (AV) enhancement, but it remains unclear whether age and cognition impact AV integration. The first aim of the current study is to investigate how age and cognitive factors affect AV integration when listeners are exposed to both natural and degraded visual speech signals. The second aim is to examine whether the neural processing of continuous speech is modulated by AV integration and ageing. Behavioural and EEG experiments (N=33, age range: 20-72) were used for these purposes respectively. Using the AV GRID corpus, two degraded versions of the videos were presented to the participants in addition to the unprocessed (natural) version. The target speaker was presented in a two-voice babble at an SNR corresponding to the speech reception threshold (SRT) of each participant. The subjects also completed several cognitive tasks targeting working memory (digit span and Corsi task), sustained attention (reaction time task), and inhibition (Stroop task).

The behavioural results demonstrated a gradual improvement of speech comprehension with the addition of details in the visual signals. SRTs correlated with age, working memory, and Stroop scores, but AV enhancements were unrelated to these factors, suggesting that differences between A and AV conditions were not driven by age or cognition. This indicates that AV integration for speech remains stable and robust across the lifespan.

The EEG data was used to reconstruct the target speech envelope of continuous speech. Focusing on the Delta band, reconstruction accuracies demonstrated a clear benefit of AV integration in the tracking of speech in all visual conditions. Additionally, AV reconstruction scores in the natural condition correlated positively with age, demonstrating a potential deficit of neural inhibition in older age. To conclude, AV benefit was observed behaviourally and in the EEG data for natural and degraded speech. The behavioural data showed that more visual details led to stronger enhancement regardless of age or cognitive abilities. EEG results showed improved neural speech tracking through AV integration but possible age-related deficits in neural inhibition.