I've been fascinated by spectrum analysis since I can't remember when. It wasn't until taking engineering mathematics on my way to an electrical engineering degree that I really began to understand just how interesting it really was. No need to glaze your eyes over with a discussion of Fast Fourier Transforms, but the technology now exists to do on a home PC what it used to take very expensive test equipment to do.
Couple that with my involvement in SCCA club racing as chief of sound control for the Washington D. C. Region of SCCA and you get someone who will try to answer the question, 'Why do those darn Mazdas sound so bad?' Well, here's some insight into the question. Below are some sound spectrums of a few race cars. I'll try to walk through why they sound different.
First, an explanation of what's in the pictures below. Each picture represents the sound spread out over frequency and time. To record these, I took my cassette recorder to the track and set up a professional microphone next to the track and recorded the cars as they went by. I then fed the recordings into my computer's sound board to capture a few seconds of the car passing. Then I used an analysis program from Pioneer Hill Software called Spectra Plus to generate the spectrum plots.
Along the horizontal axis is time. Think of it as the car passing from left to right (which by coincidence is just he way they went by on the track!). Along the vertical scale is the frequency with the lower frequencies (like the bass on your stereo) on the bottom and the higher frequencies (treble) at the top. The relative intensity of each frequency is indicated by the color, like the radar images of a rain storm on the TV weather reports. The scale is on the right, but suffice to say that the red is the highest intensity. It's a relative scale that is adjusted to make the spectrums show up well, not all of the cars were equally loud.
Now let's start looking at some pictures. First is a Formula Atlantic. This is an open wheel, single seat race car that is a stepping stone to Indy cars. It uses a two liter four cylinder engine that puts out about two hundred horsepower. People generally say that this engine sounds good. Notice that there is a strong fundamental frequency with a few quickly diminishing harmonics, or overtones in musical terms. This results in a relatively pure sounding note. For those of you with a technical background, any continuous sound can be broken down mathematically into a fundamental and a series of harmonics. For example, a square wave consists of a fundamental and odd harmonics, and sounds harsh and raspy. A sound made up of even harmonics will instead sound smooth. If you look at the frequency scale at the left, you can see that the harmonics have even multiples (but a little of both). Therefore you might expect a Formula Atlantic to sound pretty good! Notice also that there is not much significant sound energy above about four thousand Hertz (or cycles per second).

By the way, the little downward slope in the plot from left to right is a result of the Doppler shift, the change in frequency you hear as a car approaches and speeds away. The truly curious can measure the car's speed from this plot!
Now let's look at one of those Mazdas. The next picture is a recording of an RX-7 built to SCCA GT2 specifications by Kearney Motorsports. Note the number of harmonics of all sorts extending higher up in frequency than the Atlantic, and not all of these lines are harmonically related to the fundamental. People generally think that Mazdas have a raspy high pitched sound and indeed it shows up on the spectrum. Notice that sound energy goes up to almost nine thousand Hertz. Having so many spectrum lines is what we hear as an 'unpure' sound. If these individual frequency components are unrelated to each other as multiples of each other then it begins to sound like noise and not a pure tone.

Now let's take a look at some good old 'merican iron. Below is a GT1 Corvette with a nice sounding V8. Again there is a strong fundamental with a rich set of harmonics that don't go much above two thousand Hertz. Consequently, it has a low pitched sound but the rich harmonics give it a quality that just can't be described, it has to be felt!

That's all for now. I hope you got something out of this little discussion. I've enjoyed sharing it with you. My thanks to Pioneer Hill Software for the fine job in creating a truly impressive tool.
Updated 06/24/02
For more information on working in Sound Control, contact:
Jim Rose 703-742-3910 email: rosej@cox.net