By combining the technologies of targeted audio and talking heads, a perception experiment was performed. Unvoiced consonants in a vowel context produced using speech synthesis were to be identified. It was found that the talking head could eliminate some of the confusions between consonants that occurred when the face was not present. The study also gave the possibility to analyse distortions of the speech signal due to the targeted audio device.
Cite as: Svanfeldt, G., Olszewski, D. (2005) Perception experiment combining a parametric loudspeaker and a synthetic talking head. Proc. Interspeech 2005, 1721-1724, doi: 10.21437/Interspeech.2005-283
@inproceedings{svanfeldt05_interspeech,
author={Gunilla Svanfeldt and Dirk Olszewski},
title={{Perception experiment combining a parametric loudspeaker and a synthetic talking head}},
year=2005,
booktitle={Proc. Interspeech 2005},
pages={1721--1724},
doi={10.21437/Interspeech.2005-283}
}