On Mon, Apr 12, 2010 at 03:44:21PM +0200, Jörn Nettingsmeier wrote: > we have 32 bit floats, with 23 bits of mantissa and a sign, which gives > us 24 bits of precision, pretty much regardless of value range, thanks > to floats being floats. > > multiplying two floats will introduce a rounding error in the least > significant bit. that means it adds noise. at a level of -144dBFS. Not FS - it's relative to the resulting signal. Which makes it even more irrelevant. > if all those cable and bit transparency fetishists knew how reference > classical recordings are produced, the whole discussion (and large parts > of the high-end market) would go down in flames. For one, I've never seen a gold-plated XLR, nor any left-twisting-oxygen mic cables. Some mics are assembled by fair-haired Cheshire virgins, that of course makes a big difference. Ciao, -- FA O tu, che porte, correndo si ? E guerra e morte ! _______________________________________________ Linux-audio-user mailing list Linux-audio-user@xxxxxxxxxxxxxxxxxxxx http://lists.linuxaudio.org/listinfo/linux-audio-user
