On Sat, Feb 27, 2016 at 01:19:40PM +0000, Politis Archontis wrote: > > The point is that, since the direct method is equivalent to > > a decoding for the set of directions used to compute the SHT > > (which will be the set for which you have HRIR), there is > > nothing 'ideal' or special to it. It is just one specific case > > of the decoder + virtual speakers method in disguise, and one > > that is very probably using way too many speakers. > > The order of the HRTFs is limited by their spatial/angular variability, > and that would be estimated correctly by orthonormalization of the > measurement directions on the SHT stage (as you mentioned). It has > nothing to do with too many speakers etc..
What I mean with too many speakers is that it is know that using too many speakers for any given order (for normal HOA reproduction via speakers) doesn't give the best results. The decoder + virtual speaker method means you are placing the listener in the sound field as would be reproduced by the decoder and real speakers. So the direct method, if based on a large set of HRIR, can be expected to have the same problems as HAO reproduction via too many speakers. > I will try to show with some plots that there is no error in the > direct approach compared to a decoding stage. But it has to wait > for a bit... Not necessary, I'm not claiming there are any 'errors' !! All I am saying that there is no reason to assume that using the particual set of directions for which you happen to have HRIR, as speaker positions (which is what the direct method amounts to), is any better or more 'ideal' than using a classic set as would be used for real speaker reproduction. 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 [email protected] https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit account or options, view archives and so on.
