convolution tool and reverbs

[Rene Schmitz]

Hi Martin and all,

I know that there is a technique which would be not too difficult to
implement.
Its similar to raytracing (or geometrical optics), you work with "mirror"
sound sources.
The effect of a reflecting surface is modelled by a sound source placed
behind the surface. 
A simple example:

           |
       wall|     * speaker     ) mic(s)
           |        

Would be modelled:

     *           *             )     

Now do that for a more complex room, with more surfaces and computing
first, second, third, ... (cries for a recursive function) reflections,
you'd get a lot of sources which you must superimpose with the respective
delays and damping due to the energy spreading and surface reflectivity.
But this is relatively easy, all that you need is the distance of these
sources to your mics. (for omni directional mics, but applying a scaling
function for directional mics should be easy) The straight approach would
be to add several delayed and scaled audio files directly. But one could
get a room impulse file as well. However this would just model the
reflections, but not refraction of the waves on objects shorter than a
wavelength. I have no idea if the results would justify the efforts. 

Bye
 René

At 07:48 19.01.00 +0100, Martin Czech wrote:
>I thougth about that, too.
>I don't have much information about the physics involved,
>but one could try a raytracing approach (sinple minded).
>One defines the six walls of a cube (reflection properties,
>ie. filter, assuming that this does not depend on ray angle).
>Then the virtual source emitts serveral rays in all or prefered
>directions, ie. a pulse wave. These rays travel and get reflected
>like optical rays, acording to the dispersion filter.
>The virtual microphone collects all throughcomming rays
>inside a certain volume.
>
>This is totally simple minded and does not imply that 
>traveling waves should be computed.
>
>And it looks like a awfull lot of work, the debugging
>must be horrible without special tools.
>
>I better stick to baloons.
>
>My experiments with noise pulses, or decaying noise showed,
>that even such a total unphysical approach yields nice
>results. Eg. with cool edit one can draw single reflections
>or echoes at will.
>
>I guess it would be better to have a reflection generator
>that creates an impulse response according to some formula,
>or table.
>
>m.c.
>
>
>

opinionsexpressedher   | uzs159 at uni-bonn.de
einarethatofmyemploy   | http://www.uni-bonn.de/~uzs159
erandnotofmyself....   | http://members.xoom.com/Rene_Schmitz