[linux-audio-dev] more on damaging signals

Hi all, I posted some questions a while back about signals that could potentially damage PA speakers, or equipment after amplification. I'm starting to look at using a genetic programming synthesiser I'm writing for live performance: http://www.pawfal.org/index.php?page=WigWamJam This synthesiser can create arbitary functions for rendering samples, these functions are then randomly mutated and guided by the user by aesthetic choice. As you can imagine, during this process rather nasty sounds can be generated (this may not prove to be a useable system for live use without people getting annoyed and throwing things at me, but, erm, bear with me here :) Other than user driven selection based on the audible results, some automatic removal of functions is going to have to happen. From people's feedback on LAD these would include rendered sounds that have the following features: * high dc offset * very low frequency * very high frequency I'd like to avoid filtering the sounds as a post process ideally, but I guess some enveloping would be needed to prevent sudden jumps in dc at the beginning and end of the samples. I'd also automatically remove functions that result in silent samples. Any thoughts? cheers, dave www.pawfal.org/nebogeo