For me there is great excitement in starting off on a new venture, even when the final goal is an daunting as trying to develop a "machine" pianist to win a major piano playing competition. I have long had an interest in artificial intelligence, especially as it relates to the arts, and I feel fortunate to have been at the first Rencon conference even though it was very small. I think as word spreads about Rencon there will be many more people, especially young programmers who have also studied music, who will want to participate and develop software. I remember how much of the important early AI work was done by students working on their doctorates (who have since become the recognized masters -- some now retired). I think some students will find AI performance rendering strategies to be a great PhD dissertation topic.
I think we should not fail to overlook some of the "fallout" from the Rencon project. On one hand it seems like a novelty: create a program which can interpret piano music as well as a professional pianist. Yet in the process of trying to get a computer to simulate that skill, we can learn quite a bit about the process of musical interpretation and thus perhaps develop improved pedagogues for teaching interpretation to young music students. So perhaps a secondary outcome is teaching not only the machine to perform, but helping people to be better performers, too.
In listening to some of the performance programs at ICAD-Rencon, I noticed that several tried to show expression by the use of loudness variations and tempo retards, etc. Yet the programs didn't quite get these things right. The variations were often exaggerated. Yet it made me ask "how did I know that." When and how did I exactly learn how much expression should be used for a particular piece? The answer is not all that easy. Some of it was learned by listening to other performers. But how about learning a score I had never seen or heard before? Somehow I have a sense of what is "right" but the process of acquiring that was certainly a complex one. Thus coding that skill for a computer is ultimately going to be a difficult task. But the journey of getting there sounds fun. Alas, I probably won't still be alive in 2050 when Rencon projects a "winning" program, but in some ways I like the early stages of research the best, when there are a small number of researchers and they all know each other, etc. I look fondly back on the early days of the personal computer when everyone wrote their own software and build their own hardware peripherals and wrote to each other.
Currently I am President of ICAD, am Editor of the Newsletter of the Society for Electro-Acoustic Music in the U.S. and am Chair of the New York city section of the Audio Engineering Society. On the visual side I am also Vice-President of the Museum for the Preservation of Illustrative Art, an organization which digitizes the original art used in old magazine illustrations for preservation as a visual archive of the social history of the United States. I also teach at Dutchess Community College, a part of the State University of New York, focusing on courses in recording arts, electronic music composition, and digital color photographic reproduction.
As you can see, I have a very full schedule but would still like to continue some association with Rencon. I think you said you are planning another Rencon conference in connection with the International Joint Conference on Artificial Intelligence to be held in Acapulco, Mexico, August 9-15, 2003. It is very easy for Americans to get to Acapulco and I am free during those dates, so if there is another Rencon meeting at that time, I want to be sure to attend.
Also, you might be interested to know that the evening before ICAD-Rencon I ate Rencon, the vegetable, for the first time. It tasted very good.
Eric Somers
Professor