Auditory and audiovisual frequency band importance functions for isolated wordsa)

Frequency band importance weights were estimated for isolated consonant-vowel-consonant words presented in auditory-only and audiovisual modalities. Acoustic speech stimuli were filtered into six one-octave bands centered at 250, 500, 1000, 2000, 4000, and 8000 Hz. Twenty-one adults with typical hearing and vision completed a word repetition task in both modalities with two bands presented on each trial. Generalized linear mixed models with logistic link functions were used to estimate the weight of each band and the effect of modality on weight. In both modalities, band importance was highest for the 2000 Hz band and progressively declined for bands further from 2000 Hz. There were significant band by modality interactions for the four highest frequency bands, reflecting a decrease in weighting of high-frequency acoustics in audiovisual conditions. Exploratory analyses of phoneme-level accuracy revealed that visual speech cues impacted frequency importance weighting for both consonant and vowel perception. Study results are consistent with the idea that the bands redundant with visual cues have relatively less impact on audiovisual speech intelligibility than auditory-only speech intelligibility and suggest that auditory-only assessments of band importance fail to capture differences in band importance that occur when visual speech cues are available to support speech recognition.