Interaction of Audition and Vision in the Recognition of Oral Speech Stimuli

Abstract

Audio-visual observation of spoken spondaic words was found to be superior to recognition via audition-only under a wide range of S/N conditions. Data from five subjects supported the notion that observers rely increasingly more on visual cues for speech information as S/N ratio is degraded. Audition-only performance was found to be less variable among subjects than was audio-visual recognition. Increased variability in audio-visual scores at poorer S/N ratios was attributed to differences in lip-reading skill among untrained subjects. Speech levels so low that recognition by audition-only approximated chance behavior were found, nevertheless, to systematically improve observers' audio-visual scores as a function of increasing S/N ratio.