Play it again RAM Computers are developing a discerning ear for music
To the human ear, the robust tenor sax sound of Sonny Rollins is not easily mistaken for the ethereal quality of a flute. But the differences in sound between woodwind instruments can be quite subtle.    Now Judy Brown of Wellesley College Massachusetts, has trained a computer to distinguish one woodwind instrument from another. Her system can correctly identity the instrument about 80 per cent of the time—which is about as good as trained musicians..    One important future will be to allow internet search engines to find music that features particular instruments. (New Scientist, 17 March 2001 p 34). "This a very difficult task," she says, "It is at least five years away."    Working at the Massachusetts Institute of Technology, Brown and her colleagues at the Paris-based Institute for Musical and Acoustic analysed the timbre-- or sound quality—of the clarinet, flute, oboe, and saxophone.    Just as a room can resonate at a certain frequency, each of these instruments has its own pattern of resonance. This resonance, which is a function of the instrument's construction, amplities certain frequencies giving the instrument an "acoustic fingerprint". Bothe the flute and the oboe, for example, tend to amplify frequencies at around 1000 Hertz, says Brown. But the oboe has an additional bump at 1200 Hertz, which gives it its nasal quality.	Very little research has been done to develop ways of recognising musical instruments automatically, says Brown. So the researchers drew on statistical techniques that were originally developed to recognise different voices.    One of the most promising techniques is to subtract the effect of loud and soft notes, and then examine the response of the instrument over the audible spectrum. This reveals the instrumnent's acoustic fingerprint.    The researchers then selected 25 sound samples from commercial recordings for each of the four instruments being played unaccompanied. They then used statistical pattern-recognition software to compare the acoustic fingerprints from these samples with other recordings of solo woodwind instruments. The computer was able to match the instruments in these unknown recordings with one of the four known instruments with about 80 per cent accuracy.    In separate tests, the researchers asked 15 musicians to listen to many of the same recordings. On average they identified 85 per cent of the instruments correctly. Brown says that musicians should be better at identifying instruments than untrained members of the public. "I think computers can do as well as humans in identifying woodwind instruments."Mick Hamer    More at: Journal of the Acoustical society of America (vol 109, p 1064)