14thAnnual Conference of the International Speech Communication Association

Lyon, France
August 25-29, 2013

Effect of Context, Rebinding and Noise, on Audiovisual Speech Fusion

Ganesh Attigodu Chandrashekara, Frédéric Berthommier, Olha Nahorna, Jean-Luc Schwartz

GIPSA, France

In a previous set of experiments we showed that audio-visual fusion during the McGurk effect may be modulated by context. A short context (2 to 4 syllables) composed of incoherent auditory and visual material significantly decreases the McGurk effect. We interpreted this as showing the existence of an audiovisual "binding" stage controlling the fusion process, incoherence producing "unbinding", and we also showed the existence of a "rebinding" process when an incoherent material is followed by a short coherent material. In this work we evaluate the role of acoustic noise superimposed to the context and to the rebinding material. We use either a coherent or incoherent context, followed, if incoherent, by a variable amount of coherent "rebinding" material, with two conditions, either silent or with superimposed speech-shaped noise. The McGurk target is presented with no acoustic noise. We confirm the existence of unbinding (lower McGurk effect with incoherent context) and rebinding (the McGurk effect is recovered with coherent rebinding). Noise uniformly increases the rate of McGurk responses compared to the silent condition. We conclude on the role of audiovisual coherence and noise in the binding process, in the framework of audiovisual speech scene analysis and the cocktail party effect.

Full Paper

Bibliographic reference.  Chandrashekara, Ganesh Attigodu / Berthommier, Frédéric / Nahorna, Olha / Schwartz, Jean-Luc (2013): "Effect of context, rebinding and noise, on audiovisual speech fusion", In INTERSPEECH-2013, 1643-1647.