How invented DSP convolution?

July 7, 2005 2:27 PM

Anyone familar with DSP knows about convolution, even if it's just related to it's uses for reverb. It seems like it's slowly being used for many different purposes now from reverb to effects processing. Well, one of my teachers claims to have invented the use of convolution for audio in his thesis and was approached by Yamaha (I think it was) to use it in one of their products. I have been looking, but can't find, who discovered the audio use of convolution? What was is it's first commerical use? I believe him, but I want to confirm it, so I won't mention his name . . .
posted by klik99 to technology (8 comments total)

Convolutions have been used in signal processing for as long as the field has existed; it might be hard to pinpoint the first time someone applied convolutions to audio, or to digital audio, or whatever.

It would be easier to look up your teacher's thesis (this should be simple, given his name and alma mater), and see exactly what new results it contains. For example, he might have created algorithms that made convolution of real-time audio practical for the first time. (This page from a Google cache of Yamaha's site suggests that the company was involved in this sort of research in the 1980s and 90s.)
posted by mbrubeck at 2:44 PM on July 7, 2005

Well I'll let the cat out of the bag - I looked up his biography and resume, and found this where it says his dissertation from 1985 was called "The Transformation of Speech into Music: A Musical
Exploration and Interpretation of Two Recent Digital Filtering Techniques"
Oh well, I wanted to believe it, but I think he's just a bit crazy sometimes.
He's such an enigma - why claim something like that?
posted by klik99 at 2:53 PM on July 7, 2005

Well, based on his resume, he was in the right field at the right time to have worked on the first real-time audio convolution processors. I wouldn't totally dismiss his claim, but it does sound unlikely to have come out of that dissertation...
posted by mbrubeck at 3:12 PM on July 7, 2005

what he's saying is way too vague/general. he might have been involved in some implementation, but the general idea of using convolution in signal processing is so basic that it would be like claiming to have have been responsible for using fourier transforms (or "numbers", for a more extreme example).
posted by andrew cooke at 3:17 PM on July 7, 2005

Andrew Cooke is right; convolutions precede anything this guy might have done by a considerable amount, and were already a standard part of auditory signal processing long before even the compact disc format was invented in the 1970s (I mention this because the Philips researchers drew heavily on signal processing concepts in laying out the concept, so not even the limited claim that he was first to apply it to music is true).
posted by Goedel at 4:15 PM on July 7, 2005

Before you spoiled it, I was going to guess Dr. John Chowning, of Stanford U. He developed FM synthesis that was (along with other developments) purchased from him by Yamaha.
posted by sourwookie at 4:24 PM on July 7, 2005

Incidently, he and Dr. John Chowning are good friends - and that wasn't something he said (as it may appear he just makes huge unsubstantiated claims) but rather from Dr. John Chowning himself.
posted by klik99 at 4:35 PM on July 9, 2005

actually i may disagree with andrew cookes claim since fourier transforms were theorized to describe the way sound would work way before any technology would exist to do it. And what seems fundamental now still took a great mind to imagine it, even if the great mind is forgotten.
posted by klik99 at 4:42 PM on July 9, 2005