Effects of visual cue enhancement on speech intelligibility for clear and conversational speech in noise

Abstract

Visual presentation of a speaker enhances the auditory perception of speech information in noisy conditions (e.g. Helfer and Freyman, 2005; Helfer, 1997). Intelligibility is also improved when a speaker adopts a clear speaking style (Smiljanic and Bradlow, 2009). The present study investigates the contributions of these two intelligibility-enhancing factors in the presence of several types of noise, which vary with respect to the degree of informational and energetic masking they impose on target speech. Specifically, it measures sentence intelligibility in the presence of 1 competing talker, 2-talker babble, 4-talker babble, and speech-shaped noise. Meaningful sentences (in clear and conversational styles) were presented to participants in each modality (audio-only; audio-visual) and in all noise conditions. Participants reported all intelligible words. Our data shows overall better intelligibility for clear speech and for AV speech relative to conversational speech and audio-only condition. However, the visual benefit associated with conversational speech is significantly greater than the visual benefit associated with clear speaking style. The relative contribution of visual influences and speech clarity to intelligibility enhancements will be discussed.