To the engineers: Anyone can help me with a small problem?

[howard112211]

I am looking for a table that displays the dominant frequencies in sounds and noises of various artificial and natural sources, all across the audible spectrum, and also above and below. I've been googling for twenty minutes now and am unable to come up with anything useful.

[badtoworse]

1. The question is too broad.

You should identify the sources for which you want data

[howard112211]

3. Ok ->

It should be something of an overview, focused mainly on the audible spectrum and slightly above and slightly below. It doesn't necessarily need to contain too many detailed values (although the more details the better). A table covering exemplary things like the human voice, the wind, cars, maybe frequency ranges of different instruments, maybe a motorboat or a chainsaw, the noise generated by a thunderstorm, and so on. Then maybe also things like dog whistles. The more different types of sources, the better.

[OneTenthofOnePercent]

6. What you are looking for is simply too broad in the sense that

most of the things you are describing are vastly multi-frequency sounds.
For example, a chainsaw has MANY frequencies. That's why engines roar and sound complex. You'll get a wide range of frequencies for things liek instruments and voices. And other things like the wind and thunderstorms are wildly dependant upon the surrounding environment (air density, shape, temperature, etc...)

You're unlikely to find all this data from one source and even if you find the data scattered in many sources you will have to look at the environment and application of that data source to see if it is comparable with your own.

What is the end-game you are shooting for (ie: why do want the data and how are planning on using it)?
This may help us help you.

[howard112211]

7. Generally, I am trying to figure out which frequencies are the most present

in everyday life, and what the sources are that generate these frequencies.

[Recursion]

12. Are you limiting to audible frequencies?

That goes from 20 Hz to 20 KHz, or 1 to 5 on a log scale.

I did a little on this in grad school for my signals class; most "ambient" noise is somewhat red (stronger in higher frequencies than lower), while "white noise" as generated is a little blue (the opposite of red); actual white noise requires infinite power to produce so we've never heard it.

[Vickers]

2. You're probably going to have to do something like that yourself.

The ones I've found are either (a) grossly under your range or (b) specific to one area (musical instruments, for example).

If you decide to develop one, I'd be glad to have a copy.

:P

[The Straight Story]

4. kicking for later when I get back (nt)

[Canuckistanian]

5. Here's one that I use

http://www.qsl.net/dl4yhf/spectra1.html

It takes a while to figure out the controls, but it has a good html help system.

It even has an "Animal Voices" setting.

[howard112211]

8. Thanks. I will definately be using that.

[notadmblnd]

9. here's someone's published paper

http://mue.music.miami.edu/thesis/robert_hartman/robert_hartman_thesis.pdf

[Lucian]

10. Are you tackling the bird deaths problem?

Trying to see if there's a frequency that's killing the birds? I got the same idea after watching The X-Files episode "Drive" the other day.

Check out "Project Seafarer" on google. It actually existed.

Oh, and it wouldn't surprise me if they were dying from Corexit fumes coming off the Gulf.

[Recursion]

11. For once, knowing how to do a Fourier transform is useful...

Do you have some mathematical software like Matlab? (If not, you can use GNU Octave, which is free.)

You can load an uncompressed PCM (that's most .wav sound files) and simply do an Discrete Fourier Transform on the audio data (for historical reasons it will be called "fft" for "Fast Fourier Transform"); that will give you a spectral breakdown of the sound. I would warn you about the effect sampling has on frequencies, but pretty much anything you do is going to be digital, so it's all the same.

On edit, I wanted to add: if you do this, you'll get an array of complex numbers (eg 3 + 2j; you may have seen it as i rather than j in high school, but in this domain it's pretty universally called "j"); the magnitude of each number will correspond to the energy of the signal at that frequency and the angle of each number will correspond to that frequency's phase.