On Mon, 2007-07-23 at 14:16 +0200, Fons Adriaensen wrote: > On Sun, Jul 22, 2007 at 01:29:19PM -0400, Paul Davis wrote: > > > 2) there are some good theoretical arguments for needing more than 32bit > > floating point resolution for a mixer > > I did a quick experiment. > > N floating point signals, all of them Gaussian distributed, summed > both as single and double precision. If the difference between the > two sums is considered 'noise', the signal/noise ratios R in dB are: > > N R > ---------------- > 16 -142.9 > 32 -140.0 > 64 -136.7 > 128 -133.8 > 256 -130.7 > 512 -128.0 > 1024 -124.7 > 2048 -121.9 > 4096 -118.8
Even for this (most simplistic) summing example, an error or noise floor at -118.8 dB would still be like comparing the sound of a deafening jet-engine close up, to that of some rustling leaves in the background (which may or may not be missing in your soundscape ...) Hardly noticeable. I also wonder ... Suppose channels 1 - 4095 were an orchestra of assorted, close miked vacuum cleaners and hair dryers, and channel 4096 was a sole bagpiper. Would you then be able to figure out wether he is playing Auld Lang Syne or not? :) I would suspect the bagpipe to be lost. With that [4096] amount of channels, we are approaching something compareable to that of a football choir at a stadium, were one individual has no impact whatsoever on the resulting sound. _______________________________________________ Linux-audio-dev mailing list [email protected] http://lists.linuxaudio.org/mailman/listinfo.cgi/linux-audio-dev
