Instead, I advocate thinking about them as sound-like: Gravitational waves encode in an aural-like manner the dynamics of the source that generates them. You can almost think of as language-like: The signal that we "hear" encodes information about its source. Our goal as theorists and (eventual) GW astronomers is to understand that encoding, and thus to map those signals we "hear" into a deeper understanding of their sources.
NB: This page is left up to make sure that old links do not break, but is not being supported anymore. My group's research page, http://gmunu.mit.edu, now hosts all information about gravitational-wave sounds, and will be kept up-to-date into the future.
The following sounds encode the signal that we would measure when two
bodies of equal mass spiral into one another. These files are
available in wav and mp3 formats. The wav format files seem not quite
right - I sometimes find that my wav player doesn't exit cleanly. The
mp3 format files appear to pause a bit before they start.
Commentary: These sounds are presented in the frequency band that corresponds to LIGO's best sensitivity. For the lighter sources (the neutron stars and the pair of 2 ½ solar mass black holes), this means that we hear the "inspiral" signal - the waves generated while the bodies are relatively far apart and orbiting around one another. We hear the frequency and amplitude "chirp" up because the bodies are spiraling towards one another, eventually being driven to merge. We do not hear the final splat itself in this case because it is not in the region of best sensitivity. (This final merger may be more accessible with future detector upgrades.)
For the heaviest source (the pair of 50 solar mass black holes), the inspiral is at low frequencies that LIGO does not hear very well. However, the final splat is in fact right in band. That "pop" we hear corresponds to the black holes' final thrashing as they settle down into a new, more massive black hole.
These sounds encode waves generated by the spiral-in of stellar mass compact bodies captured by massive black holes - for example, a 10 solar mass black hole spiraling into a million solar mass black hole. Black holes in this mass range are found in the nuclei of almost every galaxy; sources of this type are one of the key science targets for the NASA/ESA LISA mission.
These files are available at present in wav format only.
Important technical note: The frequencies of these sources (and
of LISA's best sensitivity)
are far lower than that of the human ear! (The peak
sensitivity of LIGO, by
contrast, corresponds almost exactly to human audio.) Accordingly, I
had to fudge things a bit: All frequencies are shifted by a factor of
a few thousand from the way that nature would actually present them.
Think of it as the audio equivalent of a "false color" image.
Commentary: The first two sounds illustrate the impact of the large black hole's spin upon the gravitational wave. The sound is modulated - the "buzz" you should hear in the first sound - due to the large black hole's very rapid spin. It's interesting to contrast Sounds 1 and 2 - this modulation is far weaker in Sound 2. Sound 1 also lasts much longer; this is because the black hole's spin has a very strong influence on how far the small body will inspiral before falling into the large black hole. In both of these cases, if the large black hole is one million solar masses, then the orbital radius is initial around 6 - 10 million kilometers; it decreases to a radius of a few million kilometers before the small body plunges into the massive black hole. (A million solar mass black hole would itself have a radius of about 1 ½ - 3 million kilometers.)
The third sound illustrates a very different case. The initial orbit in this case is extremely eccentric - think of a comet's orbit around the sun. Each of the pops you should hear corresponds to the smaller body passing close to the black hole and moving very rapidly. The sequence of pops gets closer together as the eccentricity shrinks. In this calculation, the eccentricity drops all the way to zero, and the final inspiral is perfectly circular. (Note, we now know that this behavior is not quite right; its manifestation here is because Sound 3 was generated using an approximation to the real laws of GW emission. Newer calculations show that the eccentricity shrinks, but is unlikely to reach zero. An updated sound will be posted here before too long.)