Subject: Re: [ecasound] Question about multichannel sound cards
From: Kai Vehmanen (k_AT_eca.cx)
Date: Fri May 16 2003 - 01:42:30 EEST
On Thu, 15 May 2003, Aaron Heller wrote:
> I'm a novice user and still trying to understand all this. Could you
> explain why using jack is better than the following for recording to
> multiple files.
> ecasound -f:32,4,96000,i -a:1,2,3,4 -i:alsaplugin \
> -a:1 -f:32,1,96000 -erc:1,1 -o:track-1.wav \
> -a:2 -f:32,1,96000 -erc:1,2 -o:track-2.wav \
> -a:3 -f:32,1,96000 -erc:1,3 -o:track-3.wav \
> -a:4 -f:32,1,96000 -erc:1,4 -o:track-4.wav
Actually there's nothing wrong with the above. Problems come up when you
want to use really high channel counts, let's say 12 or 24 input channels.
With 4 channels and the above routing approach, ecasound will use 4x4ch
worth of memory bandwidth for demultiplexing. This is not a big problem,
but 12x12ch starts to be...
So using JACK is not absolutely necessary, but it does provide a nice
solution to optimizing multichannel use scenarios. But, but, what's
importannt in the end is what works for you. If your machine can handle 24
channels with the routing approach above, then use that. If you run out of
cpu power, then you can try JACK.
> And.... I've been using the following to playback M-S recordings to
> L-R (L = M+S, R= M-S). Is there a better way.
> ecasound -c -r -b:2048 -z:db \
> -a:1 -i:m.wav -ea +100 -erc:1,2 \
> -a:2 -i:s.wav -ea:+100 -erc:1,2 -epp:0 \
> -a:3 -i:s.wav -ea:-100 -erc:1,2 -epp:100 \
> -a:1,2,3 -f:16,2 -o:alsaplugin
I guess this is currently the best way.
> Ideally, I'd like to tie the setting of -ea together, under a single
> real-time control to compute: L = kM + (1-k)S, R= kM - (1-k)S, where
I'd say writing a new chain operator (or a LADSPA plugin) for this is the
best solution (takes 'k' as its parameter).
-- http://www.eca.cx Audio software for Linux!
This archive was generated by hypermail 2b28 : Fri May 16 2003 - 01:44:14 EEST