Contributions of temporal encodings of voicing, voicelessness, fundamental frequency, and amplitude variation to audio-visual and auditory speech perception.

Auditory and audio-visual speech perception was investigated using auditory signals of invariant spectral envelope that temporally encoded the presence of voiced and voiceless excitation, variations in amplitude envelope and F0. In experiment 1, the contribution of the timing of voicing was compared in consonant identification to the additional effects of variations in F0 and the amplitude of voiced speech. In audio-visual conditions only, amplitude variation slightly increased accuracy globally and for manner features. F0 variation slightly increased overall accuracy and manner perception in auditory and audio-visual conditions. Experiment 2 examined consonant information derived from the presence and amplitude variation of voiceless speech in addition to that from voicing, F0, and voiced speech amplitude. Binary indication of voiceless excitation improved accuracy overall and for voicing and manner. The amplitude variation of voiceless speech produced only a small increment in place of articulation scores. A final experiment examined audio-visual sentence perception using encodings of voiceless excitation and amplitude variation added to a signal representing voicing and F0. There was a contribution of amplitude variation to sentence perception, but not of voiceless excitation. The timing of voiced and voiceless excitation appears to be the major temporal cues to consonant identity.