On Tue, Oct 07, 2008 at 03:33:34PM +0200, Joern Nettingsmeier wrote: > what i'm looking for is exactly the behaviour i get from a stereo > correlation meter (or the W-X one for that matter), where negative means > out-of-phase, zero means mono and positive means stereo'ish.
That is not what you get from a correlation meter. + = in-phase, - = out of phase, 0 = uncorrelated. If the input is a single frequency, it indicates the cosine of the relative phase, i.e. zero for +/- 90 degrees. Mono will indicate +1, as will any single panned signal, and also a mix of independent panned signals. > my intent is to be able to tell at a glance whether a signal has lots of X, > Y, or Z information. for instance, i would expect the Y and Z meters to > linger around 0 for an open-air recording of a single musician at the same > elevation as the microphone. Z would move to 1 when a plane flies overhead > (or a subway rumbles below), and Y would indicate sounds coming from off > the median plane. The problem is that if cx and cy are the outputs of the W/X and W/Y correlation meters, then atan2(cy, cx) is *not* the direction of a source. The reason is that they both depend only on the relative phase of X and Y to W, and not on the relative amplitudes. Any sound in the left front quadrant would produce cx = cy = +1. I am thinking about a graphic display of sound directions for AMB, but it's a different thing. Ciao, -- FA Follie! Follie! Delirio vano e' questo! _______________________________________________ Linux-audio-dev mailing list [email protected] http://lists.linuxaudio.org/mailman/listinfo/linux-audio-dev
