Importance of high frequency audibility on speech recognition with and without visual cues in listeners with normal hearing

Purpose: To study the impact of visual cues, speech materials and age on the frequency bandwidth necessary for optimizing speech recognition performance in listeners with normal hearing.

Method: Speech recognition abilities of adults and children with normal hearing were assessed using three speech perception tests that were low-pass (LP) filtered and presented in quiet and noise. The speech materials included the Multimodal Lexical Sentence Test (MLST) that was presented in auditory-only and auditory-visual modalities for the purpose of determining the listener's visual benefit. In addition, The University of Western Ontario Plurals Test (UWO) assessed listeners' ability to detect high frequency acoustic information (e.g., /s/ and /z/) in isolated words and The Maryland CNC test that assessed speech recognition performance using isolated single words. Speech recognition performance was calculated as percent correct and was compared across groups (children and adults), tests (MLST, UWO, and CNC) and conditions (quiet and noise).

Results: Statistical analyses revealed a number of significant findings. The effect of visual cues was significant in adults and children. The type of speech material had significant impact on the frequency bandwidth required for adults and children to optimize speech recognition performance. The children required significantly more bandwidth to optimize performance than adults across speech perception tests and conditions of quiet and noise. Adults and children required significantly more bandwidth in noise than in quiet across speech perception tests.

Conclusion: The results suggest that children and adults require significantly less bandwidth for optimizing speech recognition performance when assessed using sentence materials which provide visual cues. Children, however, showed less benefit from visual cues in the noise condition than adults. The amount of bandwidth required by both groups decreased as a function of the speech material. In other words, the more ecologically valid the speech material (e.g., sentences with visual cues versus single isolated words), the less bandwidth was required for optimizing performance. In all, the optimal bandwidth (except for the noise condition of the UWO test) is achievable with current amplification schemes.