On Wed, Apr 04, 2012 at 09:59:45AM -0700, Aaron Heller wrote: > It is very simple... simulate Blumlein pickup and playback though > speakers at +/- 45 degrees, low-pass at 800 Hz, and use > cross-correlation to estimate the ITDs, and compare to the natural > hearing case. The result is here > > http://www.ai.sri.com/ajh/ITDs_Blumlein_vs_Natural/ > > I was pleasantly surprised that such a simple experiment yield such a > clean result. > > The 800 Hz LPF is needed to get reliable results from the > crosscorrelation. As you point out, humans may use the higher > frequency information (I'd like to see a citation for that), but the > point here is that accurate ITDs are present in the ear signals in > stereo playback even though the soundfield has been sampled at a > single point in space.
This is to be expected. For the frequency range considered, and the distance between the ears of human head, the time differences amount to unambiguous phase differences. Correct reproduction of the field's pressure and gradient is all that is required to preserve these in the reconstructed field. First order does this, up to around 1/4 wavelenght from the center. At higher F it's the time difference between the envelopes of the transients at the L and R ears that matters. Clearly there's no problem for sources at +/- 45 degrees, as these will be reproduced by a single speaker. At other angles it may be interesting to consider the shape of a transient envelope in typical real sounds. I'd assume this will be some decaying exponential. Then all depends on how the brain interprets the timing of the combination of two such envelopes - one from the speaker at same side as the ear, one delayed from the other. The two will merge into a single event as the delay is very short. Ciao, -- FA A world of exhaustive, reliable metadata would be an utopia. It's also a pipe-dream, founded on self-delusion, nerd hubris and hysterically inflated market opportunities. (Cory Doctorow) _______________________________________________ Sursound mailing list Sursound@music.vt.edu https://mail.music.vt.edu/mailman/listinfo/sursound
