On Fri, Sep 17, 2010 at 3:10 PM, <fons(a)kokkinizita.net> wrote: What's wrong with running three ice1712-based soundcards (M-audio Delta 1010, Delta 1010LT, Terratec Phase88 or EWS88 MT, Terrasonic TS88) with the digital inputs setup for synchronization, via a single clock-master 1712, and multiple 1712 slaves. With Terratec, this is facilitated by their EWS-connect digital sync cable that allows digital sync without permanetly physically tying up the input, and using software to switch the source of the digital input). Once setup for synchronization, use http://www.jrigg.co.uk/linuxaudio/ice1712multi.html to setup an alsa multi creating up to 40 channels (if you can find a mobo or extender with 4 pci slots) and then reference the multi playback and capture devices in the jack configuration. Given that you can get a Terratec Phase88 on ebay for $9.99 these days, or an EWS88D (ADAT I/o and spdif I/o) for $19.99 like I just did... and you can use the money you save over RME to get better speakers/amps or even a whole new computer just for that used, no longer supported by Vista, but working nicely on linux $20.00 soundcard.... One of the issues I've been thinking about w/r/t multichannel is that you actually have to route all that audio to speakers, amplifiers, etc. Seems like it would be better, say, with a 24 channel setup, to have three ADAT cards (like the $19.99 EWS88D i recently won) and then run lightpipe to a cheap D/A like http://www.aphex.com/141.htm or even better to an ADAT->AES/EBU ( http://www.aphex.com/144.htm ) and then run AES/EBU to small powered monitors. That way you could at least limit the spidernest of 24 analog connections all going to the computer, and instead spread them out closer to their target destinations. See also: ........................... http://alsa.opensrc.org/index./TwoCardsAsOne http://www.johnlapeyre.com/linux_audio/linux_audio.html ftp://ftp.terratec.net/Audio/EWS/88MT/Manual/EWS88MT_Manual_GB.pdf The EWS®-Connect connector consists of 2 pin headers, CN1 (5-pole) and CN2 (3-pole), whereas CN1 consists of a black and a white piece. If you have installed only one EWS88 in your system, just leave everything as it is: Jumper on CN1 (black bit). In this configuration, the card receives masterclock from its own clock-generator (internal), or from the digital input(S/PDIF or ADAT) of an external clock-generator (external). You can switch between the two in the ControlPanel. When installing another EWS88, you need to connect the supplied thin cable on the “master” card to CN2 (Sync-Out). This connects to CN1 (Sync-In) on the first “slave” card. You need to keep and store the surplus jumper. More cards? No problem: Simply run another cable from CN2 (Sync-Out) of the first “slave” card to CN1 (Sync-In) on the second “slave”. And, surprise surprise, another one from CN2 of the second “slave” to CN1 on the third (and last) “slave”. The EWSâ-Connect connector also provides the plug-in point for the optional EWSâClockWork. This module will allow you to run all EWS-systems in your PC from an external WordClock- or SuperClock (256 FS)- signal in sync. For further information on this, please refer to the EWS-ClockWork manual. Niels http://nielsmayer.com PS: That's good to know that my EWS88D is capable. Now all i need is an Aphex 144 and eight 5-10W spdif-driven class-d power amps (skip analog entirely) driving small full-range loudspeakers.... or even better, totally flat "planar" speakers mounted into the walls with appropriate compensation for their acoustics and resonance.... a technique that should be applied to automotive audio where all "nonfunctional" surfaces could be turned into "speakers."....

On Sat, Sep 18, 2010 at 05:35:03PM -0700, Niels Mayer wrote: Such speakers do exist, Ircam in Paris has done quite some research on them. They are based on AMB theory, but they do not provide AMB playback - what you get is a speaker that can generate multiple steerable beams. It's used in acoustic measurements to excite just part of a space, and in EA music to simulate the directivity of real instruments. Ciao, -- FA There are three of them, and Alleline.

On Sun, Sep 19, 2010 at 06:43:00PM +0800, Simon Wise wrote: What makes this sort of thing expensive is the amount of equipment and the installation costs - lots of amps and wiring are required if you really want to control each speaker separately. What is needed to make this practical is boxes with a large number of speakers, amplifiers and some DSP built-in, and just a single connection (optical MADI or network) for the signals. Train station platforms are among the nightmare places of any acoustics engineer: difficult shape, either open air or very reverberant halls, variable and high ambient noise - all of it conspires against you... Ciao, -- FA There are three of them, and Alleline.

On 09/18/2010 06:18 AM, Niels Mayer wrote: or hunt an old rme digi or hdsp 9652 on ebay, they have 3 adat i/o (plus spdif), good for up to 26 speaker feeds, if you want to go there. but before you spend several kilobucks on that many speakers: most of the content out there is first-order, which will actually sound worse over that many speakers than, say, a dodecahedron or (as fons suggested), a cube or bi-rectangle. more power to you if you build a HOA system at home, but unless more people (i.e. other than yours truly and maybe a handful of mad scientists at orange labs) get out and do actual HOA recordings, you'd better be into contemporary electro-acoustic, big time :-D