Friday, September 27, 2002 07:03 AM
Maybe rather than try and implement your own signal processing, threshold detection etc. algorithms, you could ask the user to help you -- have them press and hold down a key on the keyboard, when they are speaking and want it to record, and release the key when they don't want it to record (and press the key again when they want to resume recording, etc.).
Some advanced live musical vocal microphone systems do something similar, to avoid feedback and letting background noise come through when someone isn't singing -- instead of needing the person running the sound system to determine when to turn the microphone on or off (or when to bring its level up, in the mix), they have a foot pedal or foot mat, onstage, and the singer has to step on the foot pedal/mat, to turn on the microphone, only when they are actively singing and want it to pick them up.