We know since quite some time that 1st order upsampling has some
limits, in general.
(Not just HarpeX.)
If you have some 2nd order mikes, just try 2nd order upsampling. (This
is what HOA-DirAC and COMPASS can do.)
(You just will have more “predominant” directions, and also some
better reverb? The advantages seem to be pretty clear.)
Best,
Stefan
----- Mensagem de Fons Adriaensen <f...@linuxaudio.org> ---------
Data: Fri, 23 Oct 2020 23:57:06 +0200
De: Fons Adriaensen <f...@linuxaudio.org>
Assunto: Re: [Sursound] ORTF-3D With Higher-order Ambisonics
Para: sursound@music.vt.edu
On Fri, Oct 23, 2020 at 03:06:15PM -0400, moskowitz wrote:
My conclusion from experimenting with Harpex to generate spaced
arrays - based on comparing it to an array of mono mics - is that
Harpex's plane wave decomposition function isn't quite ready for
prime time.
Indeed it isn't. What the Harpex algorithm tries to do is to find
an *exact* match for up to two sources in each frequency bin.
There are two problems related to this:
* If either the phases of two sources are close modulo 180 degrees,
or the two directions are close to opposite, the computation
becomes ill-conditioned. This means it amplifies small errors
and the result tends to be inaccurate.
* It doesn't allow for a diffuse residual.
This is just a consequence of the maths, you can't do better
starting from just first order. Starting from second order would
solve this in many cases.
It would also allow for some more intricate cases. Imagine a
choir, spread over an 90 degrees angle, and singing unisono.
It's neither a plane wave nor diffuse, but something in between,
This is the sort of thing for which Harpex will fail, but which
could be handled by starting from higher order input.
Steven Boardman wrote:
PS don?t suppose you have a super-cardioid response of the Octamic?
We could easily provide a super-cardioid response if you can
define what exactly you mean by 'super-cardioid'. I assume it
is somewhere between cardioid and hypercardioid.
By 'cardioid' we mean the polar response that has a single
zero at the back, and the same phase in all other directions.
This corresponds to the 'in-phase' decoding in AMB terms.
By 'hypercardioid' we mean the polar response that has the
maximum directivity for the given order. In other words the
one that maximises the on-axis to diffuse ratio. This is also
the pattern that corresponds to the 'systematic' or 'max-rV'
decoding in AMB terms.
Another interesting pattern would be the one that corresponds
to the 'max-rE' decoding. It is somewhere between the two ones
referred to above, so that would be one way to define 'super-
cardioid'.
Ciao,
--
FA
_______________________________________________
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe
here, edit account or options, view archives and so on.
----- Fim da mensagem de Fons Adriaensen <f...@linuxaudio.org> -----
_______________________________________________
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit
account or options, view archives and so on.
