On Sat, Feb 27, 2016 at 12:21:30PM +0000, Politis Archontis wrote: > > No, this is not true. The decoder and convolution matrix can be combined > > into a (N+1)^2 * 2 convolution matrix. > > Ah true! by summing the terms.. > > > The only remaining difference is the set of directions. And it is known > > that using too many speakers for a given order is suboptimal. > > So what is the benefit then of adding a decoding stage in the middle?
The advantage of having an explicit decoder stage is that you can tweak the decoder for optimum results. For example it can be dual-band [1], or have some front preference, etc. Note that when doing the SHT for the direct method, you need to apply a correction matrix (depending on the set of directions) to ortho-normalise the transform. When doing that, the result is *exactly* equivalent to a systematic decoder for the set of directions used in the SHT. [1] For the direct method, you could merge the convolution matrix with shelf filters for the same result. 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.
