Natural Science Forum / Physics / Acoustics / February 2004

-

If the window is vibrating at its natural resonance you should be able to
just put your hand on it and stop it from vibrating.  If that works then
you've got your culprit.  You're theories all seem reasonable even if they
are not all likely.  As for calculating the resonance of the window, you
probably cannot.  Vibrating surfaces are very complex, try searching for
holographic interferometry and you'll see what I mean.  You could find it by
using a loudspeaker, signal generator, mic, and scope if you really cared.
You would just excite the surface with the speaker and watch for where the
response from the glass peaks on the scope.  Room modes are easy to
calculate but if that were the culprit you would notice places in the room
where you couldn't hear the sound at all and some where it was very loud, is
this the case?  While its fun to think that it could be one of these things
it may just be something in your building or near by that you simple cannot
hear during the day with all of the ambient noise.  Where I live there is
large building about 100 yards away that has air conditioners.  During the
day I cannot hear them but at night when it is quiet, it sounds like they
are in the same room as me.  The range of amplitude that human hearing is
sensitive to is very large, that's why we use a logarithmic scale (decibels)
to measure sound intensity.  We are able to listen to relatively large
amplitude sounds, cars, dogs barking, etc. during the day and not think much
of their level and then hear far more subtle things when the loud
distractions are gone.  If you can't track down the culprit then I would
suggest getting one of those white noise machines that they sell at places
like sharper image.  You could set it to a very low level and it would drown
out the annoying sound so you could sleep.

Hi,

It appears likely to me that there is a LF sound source somewhere
nearby you, and that it excites a room resonance in your room. Opening
the window changes the mode pattern, and possibly lowers the Q (the
"peakedness") of the resonance.

Attach a microphone to your computer and run a spectrum analysis
program on the computer. You could try my RTSect
(http://www.speech.kth.se/music/downloads/smptool/RTSect.exe), you
would have to decrease the bandwidth to for example 2 Hz, and reduce
sampling frequency to 11025 Hz and use a hanning window to get a good
view of LF sounds. Try to run the program when you hear the sound to
see if there is any correlation with your experience of the sound.
Most microphones would work, possibly you should avoid dynamic mics
since they can pick up electromagnetic hum, but if you have one give
it a go anyway. A peak at 60 or 50 Hz (depending on your mains
frequency) is likely NOT to be sound but electric noise (hum).