ISCA Archive Eurospeech 2001
ISCA Archive Eurospeech 2001

Fun or boring? a web-based evaluation of expressive synthesis for children

Kjell Gustafson, David House

Prosodic features were varied in four sentences synthesized using a developmental version of the Infovox 330 concatenated diphone Swedish male voice. The sentences were part of an interactive evaluation test carried out on a commercial website for a period of three months. 78 girls and 56 boys between the ages of 5 and 15 rated the sentences on a qualitative four-point scale. Results indicate that both girls and boys interpret large-scale F0 manipulations as representing a fun voice while longer durations are generally regarded as boring, especially by the boys. The results also confirm the feasibility of using a website for remote evaluation even with children.