I had the chance to meet Jean-Pierre Tubach when I was still a student at Télécom Paris in the mid-80. I had just decided to undertake my PhD on speech synthesis. Who could have been a better mentor? Jean-Pierre was a sunny personality, very generous and benevolent to young explorers, PhD students; he was very accessible and put people at ease right away, I have never seen the door of his office closed. He took the time to guide me through my first readings and he helped me to dispel the legitimate fear that every student feels when he takes his first steps into the world of research.

Jean-Pierre was already an authority on speech processing in the French community. The authority of Jean-Pierre rested on a brilliant scientific career. Jean-Pierre was a pioneer of speech processing in France, one of the first to tackle the difficult problem of speech recognition, fearlessly venturing into uncharted territory. It was a pioneering time, there was much to do, and all avenues were open. He was an engineer at heart, but he was also very interested in phonetics, speech production, speech perception, in other words, the 'basic' sciences behind speech communication engineering.

After his PhD, he spent more than 15 years in the IBM France Scientific Center, a real talent incubator, a kind of Google Lab of the XXth century, the kind of place that made the most talented young engineers and researchers fantasize in the 70s and 80s. The real deep appeal he had was nevertheless for more fundamental research and after his 15 years at IBM he came back to Telecom Paris to pursue his research.

With Louis-Jean Boe, he has produced in 1986 a phonetic dictionary of the spoken French, de “A à Zut”, a Benedictine work based on the segmentation and the precise annotation of more than 10 hours of speech, colossal work at the time, the connoisseurs will appreciate the tour de force.

Jean-Pierre had kept both feet on the ground; he guarded against excessive academicism, ethereal thoughts, and grand theoretical conjectures. On the contrary, he had a penchant for things that 'worked', things that could be demonstrated and that brought solutions to real problems. Collecting phonetic knowledge was a step to feed Jean-Pierre's baby of that time, the development of an object oriented language designed to mix expert and procedural knowledge for speech recognition.

He also believed a lot in the development of spoken dialogue systems. In the mid-80s, none of the technologies such as speech recognition, speech synthesis, or dialog management were very mature, but it took more to stop the intrepid Jean-Pierre! He spent at most 2 years patiently developing a small dialog system 'from scratch'... It was a time when programming in Pascal was considered something fancy - the bravest used ADA -, when machines were called VAX, when people did not hesitate to develop a Prolog interpreter in Pascal, starting almost from scratch... The price to pay was, of course, very high! Jean-Pierre and his students spent countless hours in front of the VT100 to make it work!

When I came back to Télécom Paris in 1990 as an assistant professor, I had the chance to co-supervise several students with him. The work meetings were privileged moments: Jean-Pierre was very curious, always willing to understand the details, he liked to go to the end of things, he was very demanding but was very benevolent and generous with students whom he always defended and encouraged.

The end of the 1980s and the beginning of the 1990s were a time of very rapid development of research at Télécom Paris. Jean-Pierre, in his own unconventional way, was a driving force. He had kept a form of ingenuity that was not devoid of malice, which made it very rare to find oneself in conflict with him. And if he sensed any tension, he defused it very quickly with one of those verbal pirouettes which was his trademark. That didn't stop him from throwing a few tantrums, because he didn't like to be messed with and he made it known. But like a rainstorm, it would fall hard but it never lasted very long.

Attending a meeting with Jean-Pierre was always a moment of happiness, he spoke little, but listened a lot. He knew how to cultivate this side of 'wisdom' which suited him so well and which basically allowed him to advance his own agenda without pain because you would not even notice he was in the driver seat.

We gradually parted ways in the late 1990s, when Jean-Pierre merged into the technostructure. Jean-Pierre became first head of the lab and then scientific director of the school, an extremely important role as the lab had grown very large in the early 2000s. These were the glory years of the LTCI (Laboratoire de Traitement et Communication de l’Information), which was then one of the largest French laboratories in ICT. I understand that Jean-Pierre had no particular desire for this kind of responsibility, but he was by no means a person who wanted to give up.

Jean-Pierre was an extraordinary person in the true sense of the word. He had his own style, a real originality in his way of being and behaving He was open to others, generous and everyone loved him for his simple way of being, his mischief and his kindness. I know that I’ll miss him.

Professor Eric Moulines,

Ecole polytechnique

Membre de l'académie française des sciences.