Timbre
Supplement for chapter 24
Sound files:
From Auditory Demonstrations CD (see discussion here):
-
Effect of spectrum on timbre
-
Effect of tone envelope on timbre
-
Change in Timbre with transposition
-
Tones and tuning with stretched partials (see discussion chapter 26)
​
From Why You Hear What You Hear....
What is timbre?
The American National Standards Institute defines timbre:
​
Timbre is that attribute of auditory sensation in terms of which a listener can judge two sounds similarly presented and having the same loudness and pitch as dissimilar.
​
The statement above is not too helpful; it basically says that timbre is not loudness and it is not pitch, but for all you know from this much is that timbre is a third thing that might also have a simple scale from low to high, like loudness and pitch do. In fact timbre is multidimensional. Timber is unlike pitch or loudness in that there is no one-dimensional scale (like frequency for example) that it can be mapped onto. It has been conjectured that timbre is 37 dimensional (color perception is three dimensional: the amount of red, blue, an green). 37 is the number of independent critical bandwidths on the basilar membrane. That is, in any sound there are excitations of 37 separate regions on the basilar membrane by specified amounts, and timbre would be determined by this pattern of loudness in these 37 regions. In fact only perhaps the lowest 15 or 20 regions play a large role in pitch; extremely high frequencies are less important.
​
Other things that matter: 1) Is the sound periodic? 2) Doe the envelope of the sound fluctuate, or is it constant? What the preceding sounds are like? Is the sound ramping up or down in loudness (another envelope issue)?
Listen to this sawtooth wave below
​
This soundfile is from a famous section of Rhapsody in Blue played forward, and part of it played backward. The spectrum is the same, the loudness is the same, but the part of the backwards segment sounds like gravel being dumped from a truck.
Change Ringing applet. Although you can just try out the change ringing as intended, you can also leave only one bell active (by unchecking the other 7) and then modify the timbre of that bell by changing its frequencies. Because this applet emulates a bell strike (that is adjustable under "sharp" vs. "dull" - in real life, this is accomplished by the hammer characteristics), and you can change the individual partial frequencies, it is quite nice for timbre of bells. Read the help file, and don't forget to lodge your changes by clicking on "Apply".
The sound can be captured on your computer (preferably by streaming capture, not through loudspeakers) and analyzed, perhaps for autocorrelation and pitch.
Recognizing musical instruments from their timbre (attack and finish removed)
Less than perfect recognition, especially for some instruments (notably flute, trumpet). Note the success with the oboe and clarinet.
Pitch (with timbre) vs. chord vs. arbitrary sounds
From "Sound Structure in Music" by Robert Erickson UC Press, 1975
​
This simplification of the relations between pitched sounds with a timbre, chords (with several harmonically related pitched sounds notes (whether parsed or not by the listener inside them, and sounds with no discernible pitch or chord structure, is provocative. It is suggestive of sounds that grade between the extremes at the corners of the triangle.
Project: Phase and timbre of wind instruments
Using at least two different wind instruments, for example a saxophones and a trumpet, determine the waveform (sound trace) just inside the bell as given notes are played, e.g. a G3, by placing a microphone there and recording with Amadeus or similar program. Try to see what the variations in the waveform there are from instrument to instrument, and player to player. Use at least two different instruments and two players, but more is better. Try to match the sound traces by playing with the phases in Fourier or MAX Partials. Do you hear a timbre difference after the phases are changed?
Hereis an advanced question (even research level): why does the instrument + player system wind up choosing the phase relations it does? We will see these phase are crucial (and well understood) for a bowed violin, but less is known about the choices of phases in wind instruments. How much do they vary from player to player? From one instrument to another, even between two different trumpets, or playing loud vs. soft?
​
Project: Phase as a function of where you are standing
Record stereo speakers playing back various controlled constant tones with two different monaural microphones placed in various positions outdoors, and also in a soundspace. Compare and analyze the waveforms, in terms of the relative phases of the partials and the perceived sound: how different is it, and how different are the phases? Analyze the geometry of the situation: how different should the phases be given the wavelength of sound, and the positioning of the two speakers and the two microphones? Make sure to adjust these so that there are significant differences in some cases.
​
Project: What instrument is that?
What makes an instrument unmistakable when it is heard in play? Is it timbre alone, or other cues such as attack and finish?
Listen to "mystery1" below and try to identify the instrument or instruments being played. Their usual attach and finish has been obliterated.
In "mystery" below, the sound above is repeated, then two other instruments. What are the new instruments?
Answers here (PDF)
This example is fairly blatant; since the spectral signatures of the two instruments differ greatly (see the answer PDF). Many other examples are possible and it is certain you can think of voice or other interesting instrumental cases, some more ambiguous than others.
Project: Clinical sounds
In "ambigous" below three sounds are short, context-free extractions of much longer solo passages. One is a violin, one is a flute, one is voice. I think the "clinical" description is appropriate - do you agree? How much longer a passage makes the instrument more recognizable, and the sound less clinical?
"BellSegment" is the bell tone segment, and the whole tone, mentioned in Why You Hear What You Hear.