Dolltalk
Cati Vaucelle
(creator)
Tristan Jehan
(audio analysis/processing)
Help from Kimiko Ryokai
Based on the observation that when children enact dialogues between dolls they
move the one that is supposed to be speaking. By combining motion signals with
the voice of the child, the system knows which doll is talking. Each character'
speech is recorded and played back as soon as the child puts the doll on the
magic stage. The software additionally processes the voices by shifting the
pitch up or down to create the two character voices necessary in a story's dialogue.
Toys such as this one allow children aged 5-7 to take multiple perspectives
on their own play, encouraging them to reflect on language, story and communicating
with others. This perspective taking is essential for later literacy skills
and academic competence.
Design versions 2000, 2001, 2002
More
details on the project