Testing audio-visual familiarity effects on speech perception in noise

The current study investigated the effect of familiarity with the voice and face of a talker on a subsequent speech intelligibility task. Participants were first familiarized with four animated characters that they learned to identify by voice, by face and by both face and voice. Ceiling level performance was reached. Following this training, participants were given a speech in noise identification task. Three types of talker condition were tested: familiar voice with familiar face, familiar voice with unfamiliar face, and unfamiliar face and voice. The results showed speech perception was more accurate in the familiar voice and face condition compared to the unfamiliar face and voice condition (a talker familiarity effect). Performance in the familiar voice with unfamiliar face condition did not differ from the unfamiliar face and voice baseline. In part, these results support the proposal that talker familiarity effects arise as a product of exemplar-based processing in speech recognition.