Actually, if you strike it correctly (e.g., against your knee, with jeans on, or the palm of your hand), there is very little percussive tone.
Don’t strike a tuning fork on a hard surface. Not good for the tuning fork, not good for the hard surface, and not as good for tuning.
After striking the tuning fork, press the base of it to the bridge of your guitar, or face of your violin, for a nice amplifier. That also helps you hear the beats to tune against. When tuning a guitar, play the A string 2nd harmonic, and match that to the tuning fork, with no beats. Tune the guitar, then repeat the process.
Because it’s designed to be resonant at a single frequency (based on the length of the tines). The tines counter-oscillate in a very stable pattern. Almost all of the flexing happens in the curved part of the shape, the straight tines don’t flex much at all. Not because they’re stronger there, but because of the way the forces naturally add up.
The air has very little to do with it. That’s a good thing; the temperature, humidity, and pressure won’t affect the note.
The thing the tuning fork lacks is a good means to transmit the vibration into the air, which is why you hold it against some part of an instrument that’s designed to turn vibrations in the wood (or whatever) into vibrations in the air. Even a table works remarkably well.
Looking at the waveform on an oscilloscope is an almost useless way of determining the level of harmonic content, unless the harmonics are of really high amplitude. You can have significant low order harmonics present and the eye really can’t pick the difference between the slightly thinner, fatter or slightly rounded waveform and a pure sine. A Fourier transform on the other hand is the correct mechanism, and an FFT analyser will give you a direct readout of the harmonics.
Bowed instruments are inherently driven by a triangle wave - the bow string catches and releases the vibrating string, and so are filled with harmonics. Woodwinds are driven by a vibrating reed, which similarly is an asymmetric drive. Brass much the same, but mostly worse. So pure resonator devices like a flute wins here. Even a tuning fork will have some harmonic content. A simple analysis of the geometry will show that there are subtle non-linearities in the coupling of the forces. But as far as simple mechanical devices go, it is pretty hard to beat.
Yeah, I seem to remember learning it was because oboes are very difficult to tune, so rather than having the orchestra sit there for hours while the oboe fiddles about, they all just tune off of him.
The history of practical standard pitch, so to speak, is long and quite interesting. I’ll make a quick list of fun facts if the thread continues to drift that way.
Thank you. I’m surprised at the number of slightly-confused answers in this thread. To reiterate: looking at the shape of a waveform is nearly pointless. Something may look a lot like a sine wave, but actually contain noticeable frequency content above the fundamental.
A nitpick, though, bowed-strings are roughly modeled by a sawtooth wave, not a triangle wave.
(I’m also at a loss to understand how a sound being deficient in overtones would rule it out as a sine tone candidate: a sinusoid has no overtones.)
Anyway, I ran FFTs of some of the sounds mentioned in the thread.
Let’s start with a transverse flute. The bright horizontal lines represent the harmonics. The fewer, the closer to a “pure” tone. As you can see, the transverse flute isn’t anywhere close to a sine tone. The vertical ripples are caused by slight vibrato and tremolo.
We move on to finger on glass. There’s a very noticeable (both visually and sonically) second harmonic. There are also inharmonic components that give the sound a distinctive colour. In a purely harmonic sound, the partials (horizontal lines) should be evenly spaced, but you can see that this is not entirely the case.
There there is the ocarina. The second harmonic is very weak, but the third is comparatively strong. The fourth is also weak, and the fifth is a bit stronger. There isn’t much past that, which is consistent with the way the ocarina sounds; it’s very close to, but not quite exactly a pure tone.
Finally, a tuning fork. We see some harmonics after the fork is struck (on the left of the plot), but they quickly fade out. After that, it’s pretty much a single lonely harmonic. In other words, a pure sinusoid.
It’s my understanding that a clarinet is almost completely impure: The fundamental is extremely low-amplitude, with most of the sound coming from the overtones. But because of the way our ears and brain work, a bunch of overtones without the fundamental still sounds like the same note as the fundamental.
Ack!! :smack: Quite right. The important bit is that the fast edge of the wave is the release from the bow, and the slow edge the drag by the bow.
Another good one are very big organ pipes. Something that surprised me quite a bit. There is some energy at the fundamental, but harmonics totally dominate the sound. Some of the perception of deep bass may come from the missing fundamental effect.
I am unable to interpret your FFTs. A spectrum or acoustic response curve vs. time would mean more to me. But I question your interpretation of the flute, at least. I recall the settings on a Hammond organ when I used to play it, and a simulated flute was a “6-4”, which means a lot of the fundamental and about a 2/3 level of the octave up, but nothing more. That’s about as pure as you can get and still sound like an acoustic instrument (a fundamental only would be purer, but sound a more electronic). A brass simulation had many overtones; a clarinet, only the odd ones.
Where did you get the sound samples? Can they be relied upon to be accurate, with no coloring or distortion introduced by the recording?
They are spectra along the vertical axis, and time along the horizontal axis, with the magnitude represented by color. Frequency increases from bottom to top, and time increases from left to right.
The plots are standard short-time Fourier transforms generated using the nice (and free) package Sonic Visuzaliser, although most audio editing or analysis software will output similar plots. I used a window size of 8192 samples, which obscures some of the temporal detail but provides decent resolution on the frequency axis.
For those who are not familiar with FFTs or short-term Fourier transforms, the idea is this: the Fourier transform is a mathematical formula that allows you to take any arbitrary wave (i.e. sound) and describe it as the sum of an arbitrary number of sinusoids (sine or cosines). In other words, this means that you can create any and all sounds just by adding sine waves. It also means that the sine wave is the “purest” or simplest sound. Harmonic sounds are made up of sine waves that have frequencies that are integer ratios of some base frequency, the fundamental.
The classical Fourier transform is one of the most useful mathematical formulas ever discovered, but there’s at least one major problem: it sucks at describing how sounds change over time. The way to get around this is the short-time Fourier transform. You essentially slice the sound in overlapping snippets and perform a Fourier transform on those.
The fast Fourier transform (FFT) refers to algorithms for rapidly computing the Fourier transform of a digitally-sampled wave. The accuracy of the FFT is determined by the size of the data, which is why longer windows yield a more accurate description of the frequency content of the sound.
In the plots I made, the horizontal axis is time. The vertical axis is frequency. The amplitude of each frequency band is shown by the colour, with brighter colours louder.
As for the quality of the recordings, I think they’re good enough to answer the OP.
The Hammond “flute” sound mimics the flute stops on pipe organs rather than actual flutes. Of course, these stops are so named because they resemble flute sounds, but in this case the flutes are recorders and such, not transverse flutes.
Unfortunately, he didn’t have the slightest idea what he was talking about.
Human voices are very messy things and can’t come close to producing a clean sign wave… let alone a perfect one. Obviously, if she could produce a pure sign wave it would sound nothing like a human voice at all. If you were to take any DAW and look at the waveform of her voice you would be shocked at how little it looks like a pure tone. An FFT would likewise reveal a bit of a mess.
But that is what gives the human voice its character.
