The concept of spoken language processing dates back to early 1986, when I drafted a proposal for the Japanese National Project on Advanced Human-Machine Interface Through Spoken Language. The prevailing idea at that time was that problems of speech synthesis and speech recognition could be solved by a combination of speech signal processing and so-called natural language processing, which actually deals only with written languages. To me, however, it was apparent that speech is not just an acoustic signal but contains linguistic information that is not present in its written counterpart, so that a new concept and paradigm would be necessary to treat both the signal aspects and the symbolic aspects of speech as a single subject of research. It was also apparent that developments of this new field could be achieved only through close collaboration of science and technology of spoken language processing, not only by machines, but also by humans. The proposal was approved as one of the first projects to be supported by the Grant-in-Aid for Scientific Research on Priority Areas from the Ministry of Education of Japan.
From the beginning, the project was publicized internationally. I was invited to give a keynote on the project at the European Conference on Speech Technology (a precursor to EUROSPEECH) held in Edinburgh in September 1987, and also was invited to give a talk at the DARPA workshop on Speech held in Pittsburgh in June 1988. The idea of spoken language processing was adopted by DARPA, which immediately re-defined one of its research goals to be Spoken Language Systems. It was then widely accepted in the speech research community worldwide, and a number of projects and research laboratories were either started or re-named to include “spoken language,” “spoken language processing,” or “spoken language systems” in their names.
Although the project was domestic and involved only Japanese researchers, it was apparent that the mutual benefit would be great if we shared ideas and results with research projects that were going on or being planned both in the United States (the DARPA project, etc.) and in Europe (the ESPRIT projects). Therefore, we held annual symposia during the fiscal years 1987, 1988, and 1989, where foreign experts were invited to present their own work and give comments on our project. Also, at the second joint meeting of the Acoustical Society of Japan and the Acoustical Society of America held in Honolulu, Hawaii in November/December 1988, for which I served as the cochair, I organized a special session on international perspectives of Spoken Language Processing by the participation of researchers and leaders of large scale projects in the world. This was followed by a symposium in Makaha, Hawaii, where noted scholars and leaders including Gunnar Fant, Ken Stevens, John Laver, Frank Fallside, Joseph Mariani, Louis Pols, P.V.S. Rao and many others participated and discussed long-term strategies.
Through these contacts I also became aware that our European colleagues were forming ESCA and planning to create a biennial European Conference on Speech Technology (EUROSPEECH), to be held in Europe in the fall of odd-numbered years. A very friendly agreement was reached with the European colleagues, especially with Joseph Mariani and Louis Pols, that we would make ICSLP and EUROSPEECH complement each other at least for the first ten years, and then try to merge the two series gradually into one. Thus the first ICSLP was held in Kobe, Japan in November 1990, with the enthusiastic support and participation of many researchers and leaders from all over the world, not only in speech technology but also in speech science, as well as in psychology, pathology and pedagogy related to spoken language.
The Permanent Council for the Organization of ICSLPs (PC-ICSLP) was formed in 1992 at the second ICSLP held in Banff to oversee the smooth continuation and development of the series of ICSLP. It consisted of leaders of SLP from all over the world. It was decided by the PC-ICSLP to give the sixth ICSLP held in Beijing a second acronym “INTERSPEECH 2000.” This was followed by EUROSPEECH 2001, ICSLP 2002 and EUROSPEECH 2003, and the two series were eventually merged into one in 2004, to be run by ISCA under the common acronym of INTERSPEECH.
Hiroya Fujisaki, Founder and Chair of ICSLP 1990
The ESCA enterprise
The story actually starts back in May 1987, in Aarhus, Denmark, at the conference on speech technology organized at Jutland Telephone by the Danish Teletechnical society, and supported by the European Commission. At that time, Jan Roukens, from the commission, after hearing the reports on the various activities in speech processing in European countries, suggested the idea of starting a European Association on Speech Communication, or possibly more generally on Information Processing.
Some months later, in September 1987, we were in Edinburgh for the “European Conference on Speech Technology” organized by John Laver, and we spoke again about this European Association. A Steering Committee was formed, chaired by René Carré, and including Vagn Wissing from Jutland Telephone, John Laver, Jens Blauert, Louis Pols, Didier Bouis from the EC, Max Wajskop, who was editor in chief of the “Speech Communication” Journal, Jean-Pierre Tubach, president of the Speech group of the French Acoustical Society (SFA), Roger Moore, president of the speech group of the UK Institute of Acoustics (IoA), and myself. Following the success of that conference, I started studying the possibility to organize regular conference activities in Europe, after conducting an in-depth analysis of the IEEE-ICASSP conference, which estimated that the European “speech” community gathered about 3,000 people at that time.
The meeting where the European Speech Communication Association was launched was in Brussels, on February 26, 1988. René Carré had devoted a very large amount of work to prepare the statutes of the Association, which were deposited on February 23rd, at the Prefecture of Grenoble, in France. All the members of the Steering Committee appeared as founders of the Association and we had the support of the IoA and of the SFA. René Carré announced that he was unable to chair the Association, and I was elected as president of ESCA. We were joined shortly after by Bjorn Granström and Giancarlo Pirani, as new Board Members, and we were all eager to build on this European initiative.
The activities started with the organization of a biennial conference entitled “Eurospeech”. The first one was organized in Paris, in Fall 1989, with Jean-Pierre Tubach as General chairman, and I acted as Technical chairman. We created the ESCA medal and the first one was granted to Gunnar Fant (KTH, Sweden). Eurospeech’89 was a big success with more than 600 participants, while the following year, in 1990, Hiroya Fujisaki organized the first International Conference on Spoken Language Processing (ICSLP) in Japan, also embracing the complete scope of speech science and technology. We immediately agreed to alternate Eurospeech and ICSLP, each year in the Fall period.
It was also decided to organize smaller scale topic oriented workshops, called “ESCA Tutorial and Research Workshops” (ETRWs), and Roger Moore accepted to organize the very first one in Edinburgh as a satellite workshop of ICASSP’89. The topic was on the use of Neural Networks for speech processing, and it was given the nick name of “N'euroSpeech”.
Third was the launching of a Newsletter, NESCA, and here, Maxine Eskenazi, then at Limsi, accepted to act as the Editor in Chief. Martine Garnier-Rizet, who was then conducting her PhD thesis, joined to help in the day-to-day secretariat. We also needed a logo. We sent a Call for Ideas, and we got a dozen answers. The winner was the logo designed by Dominique Béroule, also a researcher at Limsi. In the ESCA logo, the signal becomes the letters of the association, as a wonderful illustration of the mystery of speech, going from a continuous signal to a sequence of meaningful discrete symbols.
Starting from nothing with no money, building on it, getting members from all over the world, increasing the financial income, winning grants from the EC to support students and former Eastern Europe researchers, organizing new events, all this was a wonderful experience, and I would like to warmly thank all those who participated in this pioneering action, now that ESCA and Eurospeech have become international.
Joseph Mariani, ESCA President 1988-1993