Does adding a synthetic face always enhance speech interfaces. http://www.stanford.edu/~nass/comm369/pdf/SyntheticFacea ndTypeOfVoice.pdf

Does a synthetic face enhance the social appeal of TTS-speech interface and recorded-human-speech interface? A 2x2 between-subjects experiment (synthetic face vs. no face by TTS vs. recorded human speech) was conducted (N = 48). Results show that the synthetic face enhanced the social appeal of the TTS-speech interface, but decreased the social appeal of the recorded-human-speech interface. A consistency argument is proposed