[linux-audio-dev] mixing algorithms...

Hey there... I have an incredibly naive question... How does one mix n channels of audio down to one channel? I've scoured the 'net as best I could and haven't really found anything very authoritative. Suppose I'm dealing with float point data between -1.0 and 1.0 and have n channels. I know physically, it's all a summation of the individual waves, but strictly, summation of multiple waves between -1.0 and 1.0 doesn't keep you between -1.0 and 1.0. Looking through sox, they tend to multiply by some gain value relative to the number of channels. (namely... an average) So is averaging considered a valid professional audio method of digital mixing? what if your source samples are all 16bit? Wouldn't you need to dither in a case like that? Doesn't averaging also imply truncation of your source signal? Am I missing something here? I've tried looking to see what jack and ardour do, but was lefting wondering WHERE to look. respectfully, d!