Sol's Song

August 10, 2015 12:53 AM

A guitarist captures visible sound waves by placing a camera inside his guitar, pointing it towards the Sun, and making the strings dance.
posted by banal retentive (17 comments total)

I don't think what you're seeing represents sound waves. The strings actually vibrate much more quickly than they seem to there, and they don't form those classic wave patterns either. The video is an optical illusion - a form of rolling shutter.
posted by cincinnatus c at 1:55 AM on August 10, 2015 [16 favorites]

Yep, this is really pretty, but definitely worth noting that it's a product of a rolling shutter effect, rather than a capture of anything acoustically significant. There's a tutorial video for it, actually, here, in case you feel like doing it.
posted by howfar at 1:58 AM on August 10, 2015 [4 favorites]

But maybe it's effectively capturing them in a representative form? I presume that the apparent motion really records the string frequency modulo the shutter frequency, which means that you could derive the actual string frequency by looking at the apparent vibrations.
posted by Joe in Australia at 2:00 AM on August 10, 2015 [2 favorites]

Representative sound waves, rolling shutter effect, whatever it may be -- this was... oddly satisfying. Thank you.
posted by Two unicycles and some duct tape at 2:21 AM on August 10, 2015 [1 favorite]

Reminds me of this... but it's a zither... the mesmerising opening main title sequence of Carol Reed's film The Third Man (1949)...
posted by Mister Bijou at 2:28 AM on August 10, 2015

The guitar thing was mentioned previously. (with more, previouslier)
Specifically, this video, with a better song.
posted by MtDewd at 3:34 AM on August 10, 2015

Joe in Australia, it depends on the rate of the sampling frequency compared to the frequency of the frequency of the samples. A signal needs to be sampled at least twice per wavelength to be reconstructed faithfully (the Nyquist criterion). I image that the oscillations of the guitar string occur more rapidly than the shutter-speed of the camera, so you won't be able to derive the string frequency.

To conceptualise that a bit more clearly, imagine sampling a sine wave once per period. In that case you'd get a point at the same height in the y axis every period, so your sample would give you a flat line. If you sample every 1.5 periods, (starting at a peak for connivance) you'd sample the first peak, miss a trough and peak, then sample the next trough etc, which would make it seem like you had a wave with three times the wavelength of the original one. From this sampled wave you've got no way of knowing the frequency of your original wave, without increasing your sample rate.
posted by Ned G at 4:23 AM on August 10, 2015 [7 favorites]

The strings do not directly make the sound so we are not looking at sound waves. It's the vibrations of the air inside the cavity that make the sound. Also we are looking at frequencies of ~ 1 hz to ~ 50 hz and sound is 100 hz to 15 000 hz and the strings are vibrating much more quickly than what we are seeing.

Still the video is very pretty! I did not listen.
posted by bukvich at 6:49 AM on August 10, 2015 [1 favorite]

It's not sound waves. The strings are quaking in fear of the wrath of James Hetfield.
posted by greatgefilte at 7:22 AM on August 10, 2015 [1 favorite]

Also I found a new audio spectrum diagram which I have no idea how accurate it is but it looks really swell.
posted by bukvich at 7:42 AM on August 10, 2015 [1 favorite]

Just to be clear, strings vibrate in modes. When you pluck an open guitar string, the mode will be the 1st harmonic, which be moving at its widest right around the 12^th fret and will be sweeping back and forth at a rate of roughly 82.4 times per second for the lowest string in standard tuning.

The string length of a standard dreadnought is about 25.3 inches with is a total wavelength of 50.6 inches, whereas for the low E string being played open, I counted roughly 5 full waves, and since the sound hole is roughly 4 inches in diameter, we can guesstimate that the 5 waves are spanning 3 inches, which is a wavelength of 1.6". So totally different.
posted by plinth at 7:43 AM on August 10, 2015 [2 favorites]

Oooooh set those strings a'trembling! I like the little, quirky things aficionados come up with as they explore potentials.
posted by Oyéah at 9:09 AM on August 10, 2015

This is a good opportunity to share my favorite sound hole video. Comfortably Numb on a lovely Lowden guitar.
posted by art.bikes at 2:57 PM on August 10, 2015 [1 favorite]

A few more points on this: this isn't a rolling shutter effect (at least not primarily); those exist in still images due to the different times at which parts of the image are exposed, as the shutter has finite speed. What's going on here is aliasing (here's an example), which occurs in moving images when the sampling frequency isn't fast enough, so some parts of your subject's behavior aren't captured - the problem is a slow frame rate not a slow shutter speed. Aliasing shouldn't lead to changes in apparent wavelength for a standing wave, only changes in apparent frequency. This is as the same object in motion is being imaged, it's just that if you miss imaging an oscillation or two, it will look like it's only moved a small amount when in fact it's gone back and forth and just ended close to where it was before. That makes the result of plinth's calculation seem a bit strange, that is until another fact is introduced: guitar strings don't vibrate in a neat back and forth manner, but instead oscillate in a complicated way in an elliptical envelope. When you're under-sampling this kind of complex wave motion, all bets are off, and it could look like anything, which is what's happened.
posted by Ned G at 3:39 AM on August 11, 2015 [1 favorite]

I thought that strings vibrated in a more complex way, like a superposition of all their harmonic modes?
posted by thelonius at 6:19 AM on August 11, 2015

This one has the additional bonus of the guitar's freefall from a plane as recorded from the same vantage point (previously).
posted by kilo hertz at 11:21 AM on August 11, 2015

Calling this "seeing the sound waves" is like calling a rainbow a "spectrum analyzer". Cool, but not actually what it claims to be.
posted by IAmBroom at 7:04 PM on August 11, 2015