[sdiy] ot: rotating speaker simulation or stupid approach

[Czech Martin]

I can record the transfer charactericstic of a loudspeaker with some
microphone in some room (the impulse response, if you like).
I can repeat that, turning the speaker a few degrees of axis.
I can do this , say for 23 steps of 15 deg to make a full circle.
I must be carefull in order to get the right time shift for
each impulse response vs. all the others.
I think the impulse response will be not too long, perhaps
some 10ms.

So I get a picture of what happens when the speaker is turned,
the reflexion pattern changes, etc.

Now I can take some signal and convolute it with those 23
inpulse responses. Then I can interpolate between those
impulse responses when doing play back, in order
to simulate all the positions between two 15 Deg
measurements. 

This could give a good approximation of a rotating
speaker, though computationaly it could be more effective.
Since all the phase information is preserved,
even the Doppler effect should be visible.
I guess that if the grid of angular position would
be fine enough, this should work perfectly.

What problems will arrise, if I do it only with
15 Deg samples?

And another question:
Has someone heard or read about a circular array of fixed speakers
in order to simulate a rotating speaker?
I mean N VCAs and N amplifiers together with N little fullrange speakers,
with some control that will stear the speakers in a circular fashion.
I think this would be a hardware implementation of the first idea,
avoiding any rotating or heavy parts.
 
??


Martin Czech