Hi All,
I’m looking to do some audio analysis that would involve taking a whole bunch of audio spectra (about 3 seconds long each) and averaging them to produce one final result. I am new with Audacity and have not found this in the manual yet (assuming that Audacity can do this spectra averaging).
The project is as follows:
I am a violin maker and what I want to do is to produce an impact (tap tone) spectrum of the instrument tapping at various parts on the top plate. This will simultaneously excite all the coupled vibration modes of the instrument (top modes, air [Helmholtz] mode, back modes, and possibly neck modes). I then want to average these individual spectra together to produce one overall spectral mapping of the instrument.
If anyone can point me in the right direction for getting started with this it would be very helpful.
Thanks
Steven
Averaging isn’t too hard… [u]Mixing[/u] of two or more tracks is done by addition (sample-by-sample for digital audio).
Gain (amplification) is multiplication (sample-by-sample for digital audio) and attenuation is division (or multiplication by a value less than 1). For example, if you want to average 3 tracks, 1/3rd is about -9.5dB* and you can use the Amplify effect to reduce the gain by -9.5dB.
To get an average you simply divide (attenuate) each track by the number of tracks before summing (mixing), or sum and then divide.
If you reduce the gain after mixing, make sure to reduce the gain before you export. Otherwise you may get clipping (overload distortion).
NOTES: When you run the Amplify effect it scans the file and defaults the whatever gain is needed to normalized the peaks to 0dB. You might want to normalize the individual tracks to 0dB or some lower volume before mixing. That should minimize the variations you get from tapping slightly harder or softer.
You’ll probably want to carefully position the instrument and microphone so the results are as repeatable as possible.
\
- dB = 20 x log(X/reference)… For example, 20log(1/3) = -9.5dB.
Adjusting the relative volume levels of different tracks does not affect Audacity’s collective frequency analysis [aka plot spectrum] : you can even mute a track and its contribution will still appear on the frequency-analysis graph 
NB: the duration of a sample will weight its importance : the longer a sample the more it will contribute to the Audacity frequency-analysis plot.