[linux-audio-dev] Looking for simple and good multichannel audio soluution.

_______________________________________________________
Skicka gratis SMS!
http://www.passagen.se

attached mail follows:

Hi all !

I am looking for simple and good multichannel audio solution, of
course based on a PC and an open source OS.

I have until now managed to output 4 synchronized channels without
causing need for synchronization of hardware units with each other.
I have also managed to create sound controlled from the keyboard
using additive sythesis and with low latancy.

I know that I can use the ".asoundrc" file to make a virtual
soundcard with for example 4 analog output channels from each card
(in this case I used CMIPCI based cards intended for 5.1 sound),
but insertion of about 3 or more sound cards in a single PC may
cause problems, like a computer that refuses to boot.
You may also need to replace the Xtal oscilators with an oscilator
that is common for all the cards.

I have found cards that is expected to work with some applications
on some operating systems to output multichannel audio in 6 or even
8 channels, but it seems like these capabilities are special cases
that assumes special encoding (not just multiplexing) of the input
data.

Am I wrong ?
Are there exceptions ?
If someone knows about a soundcard, in the "consumer electronics
segment", with support for maybe 8 (or at least more than 4) analog
channels from the same amount of INDEPENDENT digital sources
(through a sinlge multiplexing application program), then PLEASE
tell me.

It may be possible that the most cost effective way to do what I
want to do is to use a digital interface (like ADAT, which is
assumed to work over a TOS-link cable) and connect some 8-channel
D/A-convertion box or maybe, there is some PCI-based prototype
card with half-made Linux drivers and an interface that allows
an appropriate amount of 16-, 20- or 24-bit DAC circuits to
be easily connected (prefarably through digital oversampling
filters to simplify the analog parts after the hold cirquits.

An example of what I want to be able to do, is to play back 7
channels previously stored uncompressed on hard disk and at the
same time synthezise the last channel from events received over a
network card or MIDI-IN-connector with low latancy and in the same
application that is multiplexing the other 7 channels. At the same
time I may want to record maybe a couple of analog inputs. I think
that 44.1 kHz sampling rate is quite enough for me both for inputs
and outputs.
The hardware should not have too high noise levels, but still
being cheap enough to buy or build.
If I fail to observe enough realtime properties in trying this,
then I can temporarily try using RAM-disk instead of hard disk
and later redesign with proper use of multithreading or multiple
processess and priorities.

There seem to be several more possibilities, like using
several PCs with a 100Mbps or 1Gbps network card and a
sound card in every box and then using 3 analog outputs
from every box to feed 3 of the speaker amplifier channels and
the 4th analog output to output test signals to make it possible
to meassure how the soundcards are related in time and
phase with each other. Here, a problems may need to be
solved, because if playback of prerecorded audio data spans over
several sound cards (in different boxes), then the data need
to be resampled (interpolated) for a different sample rate based
on the feedback.

Is there any obvious good and cheap way for me, or another person
willing to write some code and/or using a soldering device
to implement the sound system described above ?

Best regards !
Hans Davidson

_______________________________________________________
Skicka gratis SMS!
http://www.passagen.se
Received on Sun Mar 27 08:15:05 2005