Greetings,

I would appreciate some suggestions for implementing a stand-alone
audio mixer/equalizer box using commodity components, in the shortest
development time at the least possible cost. I am tending to
want to to use four PCI sound cards in an older PIII 1u rackmount
cpu to create a four audio bus/eight inputs mixer, with
four independent equalizer channels. It should be controllable
using an arbitrary protocol over RS232 or ethernet, not require a
resident GUI, nor local mass storage except perhaps flash for program
loading, and won't do any local capture, only audio I/O at 44.1 or
48kHz sampling rate to and from unbalanced line level connections.

Some preliminary thoughts are to use a version of *IX as the o/s
and layer ALSA and LADSPA with plugins on it, connected with the
'jack' framework, and for testing just control it through remote
shell command scripts until a complete control program is written.
Since I'm using low-end hardware, I had dismissed using 'Pulseaudio'
as a framework (unless convinced otherwise).

Should I be considering some other o/s or RTOS?

Has this been done as an opensource or freeware project (I don't
find much on the 'net)?

Replies are much appreciated.

Michael

msg wrote:
I would appreciate some suggestions for implementing a stand-alone
audio mixer/equalizer box using commodity components, in the shortest
development time at the least possible cost. I am tending to
want to to use four PCI sound cards in an older PIII 1u rackmount
cpu to create a four audio bus/eight inputs mixer, with
four independent equalizer channels. It should be controllable
using an arbitrary protocol over RS232 or ethernet, not require a
resident GUI, nor local mass storage except perhaps flash for program
loading, and won't do any local capture, only audio I/O at 44.1 or
48kHz sampling rate to and from unbalanced line level connections.

Why do you want to do this rather than use an existing device that does
this? There are a bunch of devices like this out there, including a
popular one from Roland that is well-documented.

Some preliminary thoughts are to use a version of *IX as the o/s
and layer ALSA and LADSPA with plugins on it, connected with the
'jack' framework, and for testing just control it through remote
shell command scripts until a complete control program is written.
Since I'm using low-end hardware, I had dismissed using 'Pulseaudio'
as a framework (unless convinced otherwise).

Should I be considering some other o/s or RTOS?

If I were doing something like this and I were having to use generic
PC hardware, I would consider RTLINUX. It would be different if the
PC were just a front end to a dsp, though.
--scott


--
"C'est un Nagra. C'est suisse, et tres, tres precis."

msg wrote:
Greetings,

I would appreciate some suggestions for implementing a stand-alone
audio mixer/equalizer box using commodity components, in the shortest
development time at the least possible cost. I am tending to
want to to use four PCI sound cards in an older PIII 1u rackmount
cpu to create a four audio bus/eight inputs mixer, with
four independent equalizer channels. It should be controllable
using an arbitrary protocol over RS232 or ethernet, not require a
resident GUI, nor local mass storage except perhaps flash for program
loading, and won't do any local capture, only audio I/O at 44.1 or
48kHz sampling rate to and from unbalanced line level connections.

Some preliminary thoughts are to use a version of *IX as the o/s
and layer ALSA and LADSPA with plugins on it, connected with the
'jack' framework, and for testing just control it through remote
shell command scripts until a complete control program is written.
Since I'm using low-end hardware, I had dismissed using 'Pulseaudio'
as a framework (unless convinced otherwise).

Should I be considering some other o/s or RTOS?

Has this been done as an opensource or freeware project (I don't
find much on the 'net)?

Replies are much appreciated.

Michael

PCDJ will let you controll multiple outputs, not sure if it had a remote
function - but it can be controlled via keyboard short cuts and I have
seen somewhere a keyboard rs232 interface so that a dumb terminal can
be used as a keyboard on a remote machine bit like a remote KVM. There
are also KVM's via ethernet about.

thoughts about your hardware - can you get four PCI cards in a 1u rack,
is there physically enough space ?

Scott Dorsey wrote:

msg wrote:

I would appreciate some suggestions for implementing a stand-alone
audio mixer/equalizer box using commodity components, in the shortest
development time at the least possible cost. I am tending to
want to to use four PCI sound cards in an older PIII 1u rackmount
cpu to create a four audio bus/eight inputs mixer, with
four independent equalizer channels. It should be controllable
using an arbitrary protocol over RS232 or ethernet, not require a
resident GUI, nor local mass storage except perhaps flash for program
loading, and won't do any local capture, only audio I/O at 44.1 or
48kHz sampling rate to and from unbalanced line level connections.


Why do you want to do this rather than use an existing device that does
this? There are a bunch of devices like this out there, including a
popular one from Roland that is well-documented.

Acquisition cost - I need to make several of these and can't afford
commercial versions.

snip


Should I be considering some other o/s or RTOS?


If I were doing something like this and I were having to use generic
PC hardware, I would consider RTLINUX. It would be different if the
PC were just a front end to a dsp, though.

Thanks - I'll look at RTLINUX.

f825_677 wrote:

snip
thoughts about your hardware - can you get four PCI cards in a 1u rack,
is there physically enough space ?

I have a few M/B choices that permit four PCI sound devices in 1u; the first
is three PCI cards and one integrated audio chipset, the other is a low-profile
riser PCI bus with dual opposing slots (four slots).

Michael

On Thu, 20 Nov 2008 11:47:51 -0600, msg wrote:

Greetings,

I would appreciate some suggestions for implementing a stand-alone audio
mixer/equalizer box using commodity components, in the shortest
development time at the least possible cost. I am tending to want to to
use four PCI sound cards in an older PIII 1u rackmount cpu to create a
four audio bus/eight inputs mixer, with four independent equalizer
channels. It should be controllable using an arbitrary protocol over
RS232 or ethernet, not require a resident GUI, nor local mass storage
except perhaps flash for program loading, and won't do any local
capture, only audio I/O at 44.1 or 48kHz sampling rate to and from
unbalanced line level connections.

Some preliminary thoughts are to use a version of *IX as the o/s and
layer ALSA and LADSPA with plugins on it, connected with the 'jack'
framework, and for testing just control it through remote shell command
scripts until a complete control program is written. Since I'm using
low-end hardware, I had dismissed using 'Pulseaudio' as a framework
(unless convinced otherwise).

Are the four cards locked together via spdif or workclock?

I don't think jack will work nicely (easily) with four separate sound
cards. It definitely won't if they are not locked together. You could run
a separate jackd server for each card, I suppose.

I wouldn't use Pulseaudio either.


Should I be considering some other o/s or RTOS?

What kind of latency is acceptable?
What standards of reliability? (Will you lose your job if it breaks? ).

I'd use PureData with straight forward ALSA. It can interface with
virtually anything and will run headless perfectly happily. It's great
for rapid prototyping of audio ideas.

Also, have a look at Faust.

I think a Linux Music distro like 64Studio would be easier than a RTLinux
or other *nix. Otherwise you will spend all your time compiling obscure
libraries and messing about just to replicate the work others have
already done.


Has this been done as an opensource or freeware project (I don't find
much on the 'net)?

Linux-audio-users/dev mailing lists would be a good place to ask.


Replies are much appreciated.

Michael

philicorda wrote:

On Thu, 20 Nov 2008 11:47:51 -0600, msg wrote:


Greetings,

I would appreciate some suggestions for implementing a stand-alone audio
mixer/equalizer box using commodity components, in the shortest
development time at the least possible cost. I am tending to want to to
use four PCI sound cards in an older PIII 1u rackmount cpu to create a
four audio bus/eight inputs mixer, with four independent equalizer
channels. It should be controllable using an arbitrary protocol over
RS232 or ethernet, not require a resident GUI, nor local mass storage
except perhaps flash for program loading, and won't do any local
capture, only audio I/O at 44.1 or 48kHz sampling rate to and from
unbalanced line level connections.

Some preliminary thoughts are to use a version of *IX as the o/s and
layer ALSA and LADSPA with plugins on it, connected with the 'jack'
framework, and for testing just control it through remote shell command
scripts until a complete control program is written. Since I'm using
low-end hardware, I had dismissed using 'Pulseaudio' as a framework
(unless convinced otherwise).


snip

I don't think jack will work nicely (easily) with four separate sound cards. It definitely won't if they are not locked together.
You could run a separate jackd server for each card, I suppose.


Thanks for the heads-up; I had not considered creating a single virtual
sound device but only a 'jackd' per card. I found considerable discussion
of this topic on the web, for example a good overview:
http://www.sound-man.co.uk/linuxaudio/ice1712multi.html

If I combine the cards, I'll distribute one hardware clock to all of them.
In your view, what advantages are there to combining them for my application?


I wouldn't use Pulseaudio either.


Should I be considering some other o/s or RTOS?



What kind of latency is acceptable?


On the order of 10ms as I currently see it.

What standards of reliability? (Will you lose your job if it breaks? ).


Should be stable for up to four hours


I'd use PureData with straight forward ALSA. It can interface with virtually anything and will run headless perfectly happily.
It's great for rapid prototyping of audio ideas.

Interesting, sort of an audio LabView... You are suggesting controlling my
remote audio processor in an X-window?


Also, have a look at Faust.

I think a Linux Music distro like 64Studio would be easier than a RTLinux or other *nix. Otherwise you will spend all your time
compiling obscure libraries and messing about just to replicate the work others have already done.


Indeed, I am guilty of tending to do the latter. For the use of old hardware
as I intend, what kernel version would you recommend? I have preferred 2.4.27-xxx
in the past.

Thanks,

Michael

On Thu, 20 Nov 2008 12:06:34 -0600, msg wrote:

f825_677 wrote:

snip
thoughts about your hardware - can you get four PCI cards in a 1u rack,
is there physically enough space ?

I have a few M/B choices that permit four PCI sound devices in 1u; the first
is three PCI cards and one integrated audio chipset, the other is a low-profile
riser PCI bus with dual opposing slots (four slots).


Can you post a link to these options. I have had occasions in the past
where I needed to fit a "standard" PC into odd enclosures, and having
links to these type of products is always helpful.

Regards
Anton Erasmus

Anton Erasmus wrote:
On Thu, 20 Nov 2008 12:06:34 -0600, msg wrote:


f825_677 wrote:

snip

thoughts about your hardware - can you get four PCI cards in a 1u rack,
is there physically enough space ?

I have a few M/B choices that permit four PCI sound devices in 1u; the first
is three PCI cards and one integrated audio chipset, the other is a low-profile
riser PCI bus with dual opposing slots (four slots).

Can you post a link to these options. I have had occasions in the past
where I needed to fit a "standard" PC into odd enclosures, and having
links to these type of products is always helpful.


As a rabid proponent of reuse most of my projects are based upon salvaged
items; in this case I have a selection of custom mainboards and chassis from
various old IP firewalls, compression protocol accelerators, industrial
controllers, etc., which stand only a remote chance of appearing on auction
or recycle sites. If you still want data, I can poke at them for model and
ID information later on

N.B. some time back I began a list of salvage items that contained reusable
FPGAs and posted it with a few updates to the newsgroups but that effort
produced mostly disinterest.

Michael