This paper presents an extension of a talking head that was implemented to synthesize audiovisual speech utterances in German. A mapping of English to German phonemes and a subsequent refinement of the co-articulation model enables the system to generate English utterances. A modified rhyme test was carried out for evaluation in terms of intelligibility. Results show that the visualization increases the identification of isolated words from 27% to 50% when played with audio.
Bibliographic reference. Fagel, Sascha (2008): "MASSY speaks English: adaptation and evaluation of a talking head", In INTERSPEECH-2008, 2324.