[sdiy] semi ot: measurement of room impulse responses

[Czech Martin]

I have now made some effort to come to some results.

First of all I purchased a Behringer EMC8000 measurement
microphone. It is quite flat, and the S/N is as expected for
such electret capsule. I thought that the pain of the mechanical
construction, transformer isolation , capsule purchase
won't realy pay off. Next I build a preamp based
on the SSM2017.

Next I built an exitation speaker with tripod. Since I have no
moulding cutter or turning lathe this was quite laborious.
The speaker is a two way coaxial car speaker (I have recovered
from my sister's nearly fatal car accident, radio broken, one speaker
ripped of, the other one o.k., car totally crushed, ..., err and 
sister o.k.). Sometimes it pays to keep things in the cellar for 15 years...

The measurement method is cyclical noise with deconvolution.
One interesting point is speaker distortion, because it will
create spectral components that do not exist in the original
signal. The level I used yesterday seemed not to make any
problems of this sort, but this has to be investigated further.

With a 1 min signal, all noise is beautifully averaged out,
I made some clapping noise in the room, and the computer fan is certainly
not very silent, too. So far I'm very happy with the results.

There are some minor difficulties, though:
-the speaker can not reproduce signals too near to DC,
 so the noise has to be shaped in the low end
-the distortion products should not go over the Nyquist
 limit of the DAT recorder, so some high frequency cutoff
 is needed, say 20kHz droping to zero at 21kHz.
-during the deconvolution the missing low and high end
 has to be observed, avoiding division by zero or very small
 numbers
-the sample read-and-cut procedure could be more comfortable

Next steps will be:

-refine software (bandlimit noise, automatic start recognition)
-purchase lead accumulator for independend power supply
-anechoic measurement of the exitation speaker, well
 burry it (halfspace), or try in the meadow somewhere with
 absorber fibre on the ground, this will be used in order
 to filter out the speaker artefacts, like low end droop,
 or resonances etc.

The portable setup will be a CD with cyclical noise in a portable CD-player,
a portable DAT recorder, the tripod and exitation speaker, battery and car
power amp, microphone preamp and microphone tripod.
Thus I can easily travel and collect original room responses
whereever I am allowed to.

First object will be the cellar and large garage, 
there I found a very long (30 m), very narrow
corridor. And of course the staircase.
My neighbours always thought that there is something wrong with
me because of the strange noises comming from my appartment.

I they now see me with these apparatus in the cellar ...


m.c.