Fons, you've described my crude experiment with this. Left, Centre, Right, and a complete IR for 'global' as the tail. I'll have to experiment some more, and possibly truncate the 3 short irs further, but i'm getting the idea in relation to a more realistic representation, however crude it may be.<br>
<br>A question here gentlemen.<br><br>With an WXYZ setup, what do those represent? Left/Right/Front/Rear?<br>So any signal in would be a composite of position based on the strength of gain between 2 or more points?<br><br>
Is it a 'box' in effect, so any point represented would be a positional calculation based on all 4?<br><br>Or do i have this wrong?<br><br>I'm off to try out the YorkMinster conf file, and see if that gives me any listening clues.<br>
<br>Alex.<br><br><div class="gmail_quote">On Tue, Jan 20, 2009 at 9:03 PM, Fons Adriaensen <span dir="ltr"><<a href="mailto:fons@kokkinizita.net">fons@kokkinizita.net</a>></span> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div class="Ih2E3d">On Tue, Jan 20, 2009 at 06:34:27PM +0100, Jörn Nettingsmeier wrote:<br>
<br>
> to clarify things, there are no "short" or "long" IRs. what fons means<br>
> is: the early reflections are the most characteristic aspect of a room,<br>
> and they affect localisation the most.<br>
<br>
</div>I don't think I can really agree with the first part of this<br>
statement, but the second part is certainly true.<br>
<div class="Ih2E3d"><br>
> therefore, if you want to have<br>
> ultra-realistic reverb, use an IR that was measured with the speaker<br>
> where your intended instrument is and a soundfield microphone where the<br>
> listening spot is. of course, in practice this is not done.<br>
><br>
> so instead you use one reverb IR instead. it can be short (tail<br>
> truncated) to save CPU, because the tail is decorrelated (blurred) and<br>
> does not provide localisation cues, hence it would be wasteful to render<br>
> it in b-format.<br>
<br>
</div>What I suggested is:<br>
<br>
* use a short IR, containing the early reflections, for each<br>
source position, or a at least a small set of them for<br>
different areas (e.g. left, center, right),<br>
<br>
* manipulate the delay of this in function of source distance<br>
<br>
* use the same long IR, containing the reverb tail, for all<br>
sources.<br>
<br>
Of course, the real reverb tail is different for each source,<br>
but you can't hear the difference. Only its statistical<br>
properties (such as the reverb time, etc) matter.<br>
<br>
But I wouldn't say it is wasteful to render the reverb tail<br>
in B-format, on the contrary, doing that makes it very realistic.<br>
<div class="Ih2E3d"><br>
> if you can make it to lac2009, let's talk this through over pizza<br>
> (that's how i learned my first steps in ambisonics from fons, and it<br>
> works suprisingly well).<br>
<br>
</div>And pizza is easy to get here.<br>
<div class="Ih2E3d"><br>
> as i said, i would recommend against attempting such a split reverb<br>
> method, because very likely things will go haywire at some point.<br>
> better bribe fons :)<br>
<br>
</div>I have some jconv configs containing separate short IRs for<br>
different source positions, and a single reverb tail. They<br>
need some reworking (*), but when you are in Parma we can<br>
play with them.<br>
<br>
(*) For example, removing the direct sound can be a bit<br>
more complicated than just cutting it off. In most cases<br>
the LF response of the direct sound continues during the<br>
early reflection period (at very low level, you don't see<br>
it in the IR). Just cutting away e.g. the first few ms can<br>
leave enough LF energy from the direct sound to result in<br>
a bass-heavy IR.<br>
<div><div></div><div class="Wj3C7c"><br>
Ciao,<br>
<br>
--<br>
FA<br>
<br>
Laboratorio di Acustica ed Elettroacustica<br>
Parma, Italia<br>
<br>
O tu, che porte, correndo si ?<br>
E guerra e morte !<br>
</div></div></blockquote></div><br>