PressCNRS international magazine

Table of contents

Michel Pitermann

Word Doctor


© DR

Da da da—Now close your eyes—Ba ba ba.” Meeting with Michel Pitermann starts with this bizarre experiment. You watch a video of a man pronouncing very simple sounds, and you hear “da da da.” Close your eyes and now, you're convinced that he's saying: “ba ba ba.” Start over as often as you wish: you hear “da” with your eyes open, and “ba” with your eyes closed.

“I am currently working on how the integration of both visual and auditory data affects our perception of speech. You were just subjected to the “McGurk effect:” What you see influences what you hear. But for you to understand how complex this phenomenon really is, let me tell you something else. The person in the video is saying “ga,” but the original sound was replaced by “ba.” Using visual and auditory data, your brain extrapolated the sound “da” which does not correspond to any reality.”

Despite appearances, Pitermann is neither a linguist nor a biologist, but a physicist. Director of the Production and Perception of Speech team within the Speech and Language Laboratory (LPL)1 in Aix-en-Provence, he himself symbolizes the interdisciplinary nature of the laboratory. “It lets me play a pivotal role in a project at the interface of several disciplines. I can also help out research scientists who are stuck on the methodology of designing an experiment.” But his accent betrays the origins of this lover of southern France: he is Belgian after all.

“I had decided to leave Belgium even before I finished school. When studying for my PhD in Physics in Brussels in 1996, I was offered a post-doctoral position at ESA (the European Space Agency). But then I met the new director of the LPL, he was a dissertation committee member for one of my fellow students. He suggested a post-doctoral subject that I fell for instantly.” Pitermann left at the end of his post-doctoral fellowship... only to return later. In 1998, he was in Kingston, Canada, developing a biomechanical model of the face, a tool that would prove to be invaluable in research on the perception of language. “We can classify models by their degree of realism. At the lowest level, we only care about the surface. This is the option used for most animated movies. At the highest level, we model everything: bones, muscles, the fatty layer, the skin: that's what biomechanics is all about.” After a brief stop at INRIA,2  in eastern France's Lorraine region, Pitermann met up with the LPL once more, where he is now working on a new model based on his experience. While he has managed to model the bones and the thirty-odd muscle groups, he is still having problems with skin: it is “unstable.” Unless you unrealistically tack it onto the skull, when you move the body the skin starts to swell until it comes right off the screen, leaving the model looking like it was skinned alive. “Last April, thanks to research scientists at the Spoken Communication Institute in Grenoble,3 I discovered that the entire biomechanical face modeling community that used the same technique had similar problems.”

“I therefore decided to head off in a brand new direction to get this blasted epidermis under control! It was not an easy decision to make...” Demanding as it is, this model leaves Pitermann time to pursue his research on the perception of language, thanks to an international collaboration with colleagues from Canada and Brazil.


Jérôme Blanchart



Notes :

1. Laboratoire Parole et Langage (CNRS / Université Aix-Marseille-I joint lab). View web site
2. INRIA: Institut National de Recherche en Informatique et en Automatique (National Research Institute for Computing and Automation): View web site
3. Institut de la communication parlée (CNRS / Institut National Polytechnique de Grenoble (INPG) / Université Grenoble-III joint lab). View web site

Contacts :

Michel Pitermann
Université de Provence, Aix-en-Provence.


Back to homepageContactcredits