On 2012-01-12, Fons Adriaensen wrote:
Nice work ! Usually for me binaural without head tracking just produces in-the-head sound. This one is different: on headphones everything seems to be _behind_ me ! Except for the airplane at the end which appeared where on could expect it.
This is again just one measurement point, of course. The Fons-measurement, so to speak. But it's still rather interesting in at least to regards.
First, if we have all of the original data leading to this binaural solution available, somewhere, and it seems that it leads to consistently inversed perception, a simple linear, multivariate discrimination test (or if you want, a support vector machine) against the consistently frontwards dataset(s) ought to give a pretty good idea of what makes the inversion happen. Especially with more and more fine grained data sets.
I also listened on speakers, using zita-bls1 to do the conversion. This worked quite well, producing some scenes with an uncanny sense of realism.
If processed right, that is another dataset in its own right. Secondarily it a) not only gives us another data set, with different constraints (do we know those constraints as well as we should?), then b) it tells us that since most people externalize sound sources better using speakers than headphones, there is something extra happening there which we should quantify (perhaps the dynamical, head-turning aspect, but maybe not only that?), and then c) certainly that the purely binaural version might be overly specific because different people hear it in different ways (so that it should be made less specific, in a controlled way, even if there could/should be a way to make it specific to a single person's hearing, even in the fully static setting).
None of this is new in any way. But at the same time, evenwhile I'm rather well acquainted with the classical studies with dichotic (spectrally wideband and temporally low-support excitation; "analog clicks"), both temporally spread out and spectrally spread out yet temporally incoherent (bilaterally delayed white noise), the second most classical test in spectrally singular and at the same time temporally continuous power (phase discrimination tests between two sinusoids), and many other comparable tests, even in open air/headphones experiments...
I've never seen a meta-analysis being done which would really lower or upper bound the directional accuracy of human hearing. Even those few tests which purported to show that ultrasonics somehow affect directional hearing acuity, which I tentatively painted as something which could validate the highest sample rates, many years back, never did not follow such a protocol to its logical conclusion.
This sort of thing should be done, in all of its various forms, while recording and fully, openly, sharing all of the resulting data. Then all of the mass should one day be analysed en masse, using the most advanced statistical tools, to see whether/what real data could be gleaned from it, and especially which further, more precise hypotheses it suggests for further study.
-- Sampo Syreeni, aka decoy - de...@iki.fi, http://decoy.iki.fi/front +358-50-5756111, 025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2 _______________________________________________ Sursound mailing list Sursound@music.vt.edu https://mail.music.vt.edu/mailman/listinfo/sursound
