[LAU] Jconv

Fons Adriaensen fons at kokkinizita.net
Thu Mar 26 14:51:47 EDT 2009

On Tue, Mar 24, 2009 at 08:52:07PM +0300, alex stone wrote:

> It's an experiment into what happens when i use multiple instances of ER,
> and try and position them 'over' the source.
> Example
> If my stage is 200 units wide, (100L, 100 R) my 1st violins will be about
> 50L. (Just as an example.)
> That is, my dry signal in from LS is positioned at 50L.
> I add a send in Ardour, on the 1st violins track, and send this out to Jconv
> 'multi' to input 1st VL 1. (Mono in.)
> The Jconv ER 1st violins out is then brought back into Ardour, and it does
> so dead centre, at 0, neither left, nor right.
> In my experiment, i want to move that particular ER 'over' the 1st violins
> to 50L, and then, using dry and ER gains, attempt to balance between the two
> signals, and see what happens. (There's a further challenge here with depth,
> but i'm just working with L/R position at the moment.)

This does not make any sense. Early reflection are
the first 80-100ms of a room's response. There will
normally be a number of reflections in this interval,
and they come from all directions. In a good concert
hall there will be some strong ones from the sides.
Reducing them to mono and giving them the same direction
as the direct sound does not correspond to any real
acoustic process. A jconv setup meant to reproduce ER
should at least be stereo if your mix is stereo. You
just mix it directly into the stereo master bus. 

Most stereo reverbs will just produce one set of ER,
while they should be different for each source position.
If you separate the ER from the reverb tail there are
some things you can do to have more variation. One is
to create a second set by just swapping L and R, or for
an AMB reverb, by rotating it. Much more interesting
is to play with delays and levels. For a close source
the ER will be weaker and arrive later. For a more
distant source the ER will be close to the direct
sound, and stronger. 

An alternative is to use a reverb that provides a 
number of different ER patterns. The reverb I captured
in the concert hall of la Casa della Musica does 
provide this, IIRC there are at least 4 ER patterns
corresponding to different on-stage source positions.
In the jconv configuration they have separate inputs
(as does the reverb tail), and a common output. Nothing
is lost by the common output, as separate ones would
just all be mixed into the master bus anyway. You can 
enhance the set of 4 by using different delays.

On each channel you should have a post-fader send to
set the level of the ER and a second one for the reverb
tail, and insert a short variable delay into the direct
sound path (after the point where the delay send is taken).
Or you put the delay into the reverb send and use it
in the opposite sense. 

Of course doing this in Ardour is a nightmare as you end
up with three of four plugins per channel. For a typical
orchestra this means you'd have something like 50 plugin
windows on screen, or being forced to re-open and close
them for each adjustment. 

I'm developing a mixer app that has a more ergonomic
layout for this sort of thing. It wil also do ambisonics
and provide filters, dynamics, equaliser and delays on
each channel.


Io lo dico sempre: l'Italia è troppo stretta e lunga.

More information about the Linux-audio-user mailing list