Good questions all -

The accelerometer is actually a tiny piezo one to avoid adding mass the the item being measured (and thereby possibly altering the resonant frequency).

There is tremendous signal loss in air relative to soil, so while the transmitter (Tuba) must be air-coupled, we will get much stronger singals by measuring vibration directly on the earth rather than trying to get the much weaker vibrations in air.
Also, by using an acceleremoter, we avoid ambient noise.

I'm afraid I do need to worry about the Nyqusit frequency right off the bat, because it affects the "shape" (I presume you mean the spectral shape?) of the sound. If I look at the shape first, I am already looking at possibly alias-affected spectra.

The data coming in the RS-232 in raw mode is strings of 1's and 0's (duh). I do have a spec sheet that lists what each bit represents - i.e. the exact format - but I am no programmer, so I am looking for something that, if I record the strings, I can crank them through to get back decimal numbers, and store them as ASCII text, which I do know how to import to Excel.

I am probably describing the computer part poorly, becasue my knowledge lies in spectral analysis of vibrations, not computer stuff. So, once I have decimals numbers in Excel, I'm good. Until then, I have data that I don't know what to do with.

Arful Codger - I am actually interested in post-processing, not real-time conversion. I want the fastest sampling rate possible, and the real-time conversion (which can be done onboard the accelerometer) slows it down. It sounds like you have a lead on exactly what I am after, but I'm stuck in Windows! Do you know of anything that does exactly what you describe above, but for DOS or Windws? Again, I do not want to convert on-the-fly. I am happy to stream binary to some type of file through Hyperterminal, and convert later.

I really appreciate all the good tips, and am trying to ejookate myself up to your levels.