Re: Multichannel audio cards

[Matt Flax <flatmax@xxxxxxxxxxxxxxx>]

I believe in order to do 3D audio over a loud speaker array, it is
possible to use both ambisonics and more modern discrete sound source
panning techniques on the same system ....
The trick is firstly to write a software library which does the correct
sound card handling routines ....
Issues include :
a] You have to sync all the sound cards, which should not be too difficult
a task as driver developers are starting to think about this, and for
example I believe the ALSA drivers are more then capable of multi
sound card syncing with the standard release.
b] Mixer ... signals have to be dynamicly mixed down to the speakers
... ambisonics (a simple 'pinv()' routine in octave/matlab
suffices or you can see for Jot's paper [1] for more modern
approaches). Using discrete source panning (my library [2] - pdf paper
there too) or (Pulki's VBAP [3] or the Lake DSP methods [4]). [3] and [4]
are similar and face the same problems/issues.

[1] "A Comparative Study of 3-D Audio Encoding and Rendering Techniques",
Jot JM., Larcher v., Pernaux JM., AES 16'th International Conference on
Spatial Sound Reproduction, 1999.

[2] mffm 3D Audio library : http://mffm3daudio.sourceforge.net

[3] "Creating Auditory Displays with Multiple Loudspeakers Using VBAP: A
Case Study with DIVA Project", Pulkki V., Lokki T., International
Conference on Auditory Display, Glasgow, England, 1998.

[4] "Optimal 3D Speaker Panning" Dickins G., Flax M. R., McKeag A.,
McGrath D., AES 16th International Conference on Spatial Sound
Reproduction. April 10-12, 1999



On 24 October 2001, Peter Lennox wrote:
> I would strongly suggest considering a 2nd-order ambisonic setup, especially
> if you have Matlab. Derby University in the uk have a programme: 'simulink'
> which runs under matlab and which allows configuration of ambisonic decoders
> on fast PC with multichannel  soundcards.
> The benefit of this route is that ambisoncs supports periphonic
> (with-height) displays, but it is a kernel system and so is compatible with
> lower-orders of spatial encoding, so horizontal-only material is entirely
> compatible with it. It's very easy to put mono, stereo, 5.1, 7.1 , 1st-order
> ambisonics, and even binaural (though here the conversion is not perfect)
> into an ambisonic 2nd-orderdisplay, and this can be encoded in realtime with
> managable latency.
> I know they are at present experimenting with a new 32xspeaker periphonic
> setup for medium-to-large scale listening, running on a PC with 2x16ch
> soundcards (operated as a single device). I can find out the soundcard
> manufacturer if you wish
> regards,
> ppl
> ----- Original Message -----
> From: "Sanja Dupor" <106744.1511@COMPUSERVE.COM>
> To: <AUDITORY@LISTS.MCGILL.CA>
> Sent: 24 October 2001 08:16
> Subject: Multichannel audio cards
>
>
> Hi Regis,
> see EWS88MT  Terratec card at www.terratec.net
> Zlatan
>
>
> Nachricht geschrieben von Regis Rossi Alves Faria
> >
> We are looking for suggestions on multichannel audio cards (models
> and/or brands) as well as multichannel softwares (to generate more than
> 2 stereo audio outputs, be it either 5.1 or n-independent
> channels/outputs) to drive a virtual 3D (audio) environment (i.e., n
> independent speakers set in a way to produce audio spacialization).
>
> The system is to synchronize with a similar virtual 3D visual
> environment, where objects can be visualized and heard.
>
> Those who can advise on this, or think could help if having more
> information, I would be grateful in contacting.
>
> Regis Rossi A. Faria
> Media Engineering Center
> LSI - University of Sao Paulo
> Brazil
> regis@lsi.usp.br
> <

--
Matt

For electronic musicians ...
Vector Bass          : http://mffmvectorbass.sourceforge.net/
For developers ...
3D Audio Library     : http://mffm3daudiolib.sourceforge.net/
Multimedia Time Code : http://mffmtimecode.sourceforge.net/