No amount of signal processing can correct for a large brass band

Grayson got a zoom digital audio recording device today. It can record stereo at 24bits, 96kHz and appears to be quite noise free. Coincidentally Eirik and I like to go on Tuesday nights to practice organ + oboe pieces on this organ:

Also coincidentally I've been doing some work on aeolus, and was hoping to record some long duration samples of individual stops for resynthesizing. One problem is to remove the room reverb from the signal. There are many ways to do this, the simplest mathematically is to generate an acoustic impulse such as a loud hand clap, or a balloon popping. This allows one to correct for the noise generated by echos.

An alternative strategy is to play a chirp or a known noise sample (DSSS) from a speaker. You do need a very good speaker for this, or at least a sufficiently linear and powerful one that has been calibrated.

However, when we got to the church the door was open, and there was a lovely brass band sound coming out the door. It turns out that the church room was filled with a brass band. I don't have the techniques to remove such a contribution from my recording, so we decided not to play.




[æ]