|
THE MAIDO AND GURBY EXPERIMENT
The Maïdo and Gurby Experiment is a particular setup showing how autonomous creatures can coordinate socially to build a shared repertoire of syllables. There are both a scientific and a pedagogical side in this experiment. Indeed, from a scientific point of view, it is an implementation of the imitation game developed by Bart de Boer in his work on the evolution of vowel systems. Here, the imitation game is extended to syllables and to completely autonomous creatures. Agents coordinate without external supervisors to play the imitation game, i.e. to agree when to play, who will play the speaker and who will play the hearer, to perform robust turn taking, ... Agents communicate here through loudspeakers and microphones, and so their social coordination has to be robust enough to overcome the possible (human speech) sounds coming from their environment. The solution that was taken to achieve this robust social coordination had to be prelinguistic since this experiment was about the origins of speech sounds, so I could not assume a coordination scheme based on language. This is why I decided to provide them with the ability to express basic attitudes and emotions through the modulation of their voice's intonations when babbling. This is where the pedagogical side comes in. Indeed, this experiment was also made to be exhibited in La Cité des Sciences et de l'Industrie, a scientific museum in Paris. As a consequence, people passing around had to be able to understand what was happening without reading explanations. This is why I decided to build a system of emotional voice synthesis, such that humans would also directly understand the attitudes or emotions of the creatures. This led to two projects which then developped independantly: Emotional Speech Synthesis, and Emotional Speech Recognition. The scientific side of this experiment is summarized in the following paper: Oudeyer, P-Y (2005) How phonological structures can be culturally selected for learnability, Adaptive Behavior, 13(4), pp. 269--280. bibtex reference The technological side of this experiment is summarized in: Oudeyer P-Y. (2002) The production and recognition of emotions in speech: features and algorithms, International Journal in Human-Computer Studies, vol. 59/1-2, pp. 157-183, special issue on Affective Computing. bibtex reference abstract |
RELATED PROJECTS
Developmental Robotics and Artificial Curiosity
Natural human-robot interaction
|
