Thanks for all the input. I did some experimentation with jamin. Cutting of everything below 300 Hz sure helped alot. But the harmonics went right through the speech spectrum so canceling them all out pretty much meant canceling the speech out as well. I found freqtweak and hooked that up. It did produce a very beautiful spectrogram but it didn't solve my problem. I need something smarter. Something that will 1) take a few seconds of audio when there's no speech (only hum) and treat that as a baseline. 2) Reduce the frequencies all over by that baseline. In effect analyzing the tool would look at the hum and create a filter that matches it exactly. I guess there's a word for that? alex -- Alex Polite http://flosspick.org - finding the right open source
