Thursday, January 8, 2009

And I thought MIR meant peace

I am a software engineer with a specialization in signal processing and I love to play the violin. Recently I was so fortunate as to be able to bring togather those two loves - music and signal processing - into a heady mix, and have been delving into the strange beautiful amazing new field of Music Information Retrieval - MIR.

The thing is that a sharp transformation has happened here - but then that's nothing to remark, fiddlesticks, in our century. Anyways, music analysis, that somewhat esoteric field, has been around for ages. It was for theoreticians, philosophers, hobbyists, and the disinterestedly curious. Yes, there was that gentleman who claimed disinterested intellectual curiosity was the lifeblood of true civilization. But music analysis wasn't of utility to men in general. And now suddenly, its become a utility. And what's more, its become something with huge money in it. How did that happen?

Well, the simple answer seems to be: music analysis makes it possible to search for and retrieve specific kinds of music from digital databases - which includes the Internet. In its new context, as a tool for search and retrieval, its a boon for every music lover. The applications it will be put to in the near future can but be imagined. Some might be to do with automatic genre and mood recognition in music. I want to find sad songs in a voice with texture like Connie Dover's, but composed in renaissant Italy - how's that for a search query. I want to find songs in a voice that sounds like a Maggini violin.

The amount of work happening in MIR these days is quite a bit. Well, some of my first discoveries: ISMIR is the prime conference of the MIR community, and the next one will be in Kobe, Japan, in October: http://ismir2009.ismir.net/. MIREX is a competition where the submissions are software that perform MIR related tasks, like performer identification or genre classification: http://www.music-ir.org/mirex/2009/index.php/Main_Page. Curious: if four participants are guaranteed, you can create a contest at MIREX for any MIR related task of your choice. An active list discussing MIR related topics is music-ir - easy to locate the subscription info on the web.

And some papers that really got me recently (more on them in later posts) - one from ISMIR and one from far before:
- Using expressive trends for identifying violin performers http://www2.iiia.csic.es/~arcos/papers/2964.pdf
- Stochastic processes in music and art
http://www.springerlink.com/content/y5159n4316vq2w6h/

Now, if you are a musician who loves technology, or a technologist who loves music, you'd do well to go listen to the strange things those folks are saying, and check if there's something you can contribute by way of two cents or more. Seems to me MIR could do with a confluence of musicians who mayn't be tech savvy and techies who mayn't be note savvy...

My appologies to anyone of the MIR community who comes along, for being such a busybody - I hope you don't mind! Well, I do hope to be a contributor as well - in a while.

No comments:

Post a Comment

Followers

Add to Technorati Favorites