The Music Conversation. Part the

The Music Conversation. Part the Second (second part of the first part?)

I was going to go into my theory of Cultural Carbon Monoxide, but the timbre question has come up, and it is both extremely important, as well as gravely neglected in classical music theory discussion, due to a paucity of language to describe it. Read. Enjoy. Comment profusely!

In the comments to the original post on the morality of Rock and Roll music, our Accidental Choir Director raised the good point of timbre. As I mentioned in my response in the comments box, timbre is the aspect of music that we have the least developed vocabulary to use. Generally when we wish to describe timbre we resort to comparisons of known timbral sources; we might describe a sound as raspy, breathy, harp-like, etc. Or we may take a more romantic approach and describe a sound as dulcet or craggy or in some other rather abstract way. Unless we delve into the complexities of acoustics, we do not have a succinct and accurate way of describing timbre the way we can describe (and notate) pitch, duration, dynamics, or rhythm.

The example I gave of the problems of notating timbre is that of transcribing the solos of the great jazz saxophonist Pharaoh Sanders. I can transcribe a solo of his note for note, even accounting for the exquisite subtleties of jazz rhythm (although most prefer to notate jazz “straight” and mention that it is “swing rhythm” simply because the level of dotted notes and ties would be cumbersome). However, there is no way to effectively communicate the quality of his tone on each note. Without the timbral variation, the solo would lose something.

I have long proposed that a system of timbral notation be developed, but, frankly, I am at a loss as to how to do it. There is a copyright case involving the music of jazz floutist James Newton that he lost to the Beastie Boys because the sample the Beastie Boys used of his music was deemed by the court to contain too few notes to require composer royalties, simply because his intricate use of overtones was not notated as part of the composition. So there is a lot at stake here.

The big problem with timbre is that it is incredibly complex. Basically a note is built on a fundamental pitch and a series of overtones, all of which add to the quality of the sound. The overtones are part of the harmonic series and can be described in terms of number (e.g. fundamental plus the first fifth overtones). White noise (a random distribution of sound throughout the audible range)and pink noise (random distribution of sound in specific sectors of the audible range, often excluding the highs and lows) also have a place in describing timbre (for instance, a breathy tone is one in which noise is clearly audible, often with its own overlapping envelope to the envelope of the note itself – think of Stan Getz’s Brazilian recordings).

A matter of timbral consideration that comes to prominence in the twentieth century is the use of non-melodic concrete sounds (e.g. the recorded sound of a car crash), which are notoriously difficult to describe in terms of fundamental plus overtones. With concrete sounds we can call the recording itself the notation, but that does not help us develop a timbral vocabulary. Another problem that has its genesis in the twentieth century is the use of electronics to alter a sound. The engineers can be very specific in how these effects are realized, but the language of the engineer is too involved to make for useful notation. The best we can do is specify the sound source, the electronic effect unit, and the settings on the unit (for instance, we can describe a Fender Stratocaster played through a Buchla spectral processor with 20 Hz at +2, 40 Hz at +4, and so forth all the way to 20,000 Hz). Considering all the permutations of the various effects units, especially when multiple units are used, this gets to be as cumbersome as supplying a graphic notation of the wave form.

In Balinese music a system of tuning has evolved that requires that some of the instruments be tuned slightly sharp and some slightly flat. The resultant sonic interference (measurable in terms of audible beats) gives the Balinese gamelan a shimmering quality that is essential to the characteristic sound and musical interest of the music (gamelan does not have a functional harmony, or even much harmony at all beyond the perfect fifth, so this timbral variation is essential to keep the music interesting). The Balinese have taken the route of describing these effects with the language of nature, so that one of these altered pitches is called the humming bee and the other the sucker bee. Now, it is one level from the concrete, because it takes a tremendous leap of imagination to hear actual bee sounds in this. Perhaps this is the route for us to take in the West.

Until we get a developed vocabulary it is nearly impossible to have a thorough conversation on the affect of sound quality. When we describe a sound as “jarring” it can be because the sound is timbrally complex (e.g. an electric guitar with heavy distortion), because the sound is used in a dissonant harmonic setting (e.g. sine tones used in counterpoint to create a strong dissonance), because the sound is in a strange structural context (e.g. if a bassoon were introduced in a passage dominated by flutes), because the sound is dynamically surprising (the famous chord in Haydn’s so-called Surprise Symphony), or for a number of other reasons. Even to discuss a Byrd motet as being sung does not give enough information. Is this a bel canto ensemble, with the use of vibrato governed by the conventions of bel canto; is this a historically informed ensemble, using the vocal techniques we believe to have been in use in Byrd’s time; or is this an Appalachian white gospel choir singing Byrd the way they would sing “How Great Thou Art”? A famous example of this problem is in the performance of Porgy and Bess. Most any fan of this great opera would tell you that it simply does not sound right if the performers are not black. With some exceptions, we know what a “black voice” sounds like, as opposed to a “white voice.” Surely there are issues of phrasing, but even a white singer who has perfect jazz phrasing (I am thinking of Tony Bennett), is usually easily identified by voice quality as a white singer. When a Mose Allison comes around, it is a remarkable exception to the rule.

Any of the voices used in the examples above can be analyzed by various scientific means. We can make graphic notations, lengthy notes of the presence of this overtone or that overtone, or can use cruder language (often a good jazz singer is described as having a horn-like voice (Betty Carter comes to mind), which implies a certain overtone presence, but does not say it outright). However, when we are dealing with a number of voices or instruments, any analysis will become tremendously complex. I will give a crude example:

We will discuss a chord progression of ii-I (6-4)-V-V7-I in the key of C. Now, for the sake of simplicity we will presume basic three-voice voice leading (ignoring the avoidance of parallel fifths) and equal temperament, so that the first chord is D-F-A, the D becomes G, the F becomes E and the A moves to C, giving us G-E-C, the third chord being G-D-B, with the G and D remaining, but the B moving to F (ignoring the egregious melodic tritone), and finally seeking resolution in C-E-G. To a laymen this must seem complex already, but it gets even worse when we take into account timbral interactions. Let us, for the sake of simplicity, assume that all three voices have the same number of partials, namely fundamental, octave, fifth, and third. This is a bit artificial, but it will have to do.

So our first chord can be described as D (D,A,F)-F (F,C,A)-A (A,E,C). Now, the sonic energy is significantly lower in the partials than the fundamental, but we will have to speak of that relationship only generally, but we have the basic consonant third, third, fifth echoed throughout, with the D sharing the A and F with the partials of the F, the F sharing C and A with the partials of the A. The tricky partials are the C in the series that comes with F(which is a major second from D – a fairly strong dissonance) and the E in the series that is built into A (being a minor second from F, a strong dissonance). Now, anyone who plays an instrument will tell you that the sound of this chord is not dissonant, even though some of the partials are dissonant to the partials and roots of some of the other tones. However, if we are truly to analyze the sound to the level of timbre, we must speak of these, how the decreased sound energy and spacing of the partials only creates a richness to the sound without an audible dissonance, etc.

At this point we have already become bogged down and we have yet to leave this chord, let alone hit that Dominant 7th chord with the tritone built in. Multiply this geometrically if we are discussing the sound of actual musical instruments and voices with their many partials and noise elements, and you will see the depth of our problem! How tempting it is to simply jettison the whole discussion of timbral analysis and resort to “the pure flute-like tones of the sopranos with the cutting buzz of the bassoons.” Although we need to resist this temptation, because the quality of the sound clearly has an impact on the music. As our Accidental Choir Director stated, it is foolish to think of a Josquin motet without looking at how it is scored.

Trained musicians deal with these problems in concrete ways. We study orchestration and learn to combine sounds using our acoustical imaginations, and we borrow from the palette of previous composers. Some of the previous composers, for instance Ravel, have created pieces that are virtual études of timbral exploration. If I want a particular sound I look at the work of a composer that uses that sound and imitate it, combining my experience with a good bit of trial and error (less trial and error if I have more experience). But in the end, neither I, nor Ravel, nor anyone can describe these combinations in a slightly non-technical analytic way, and our musical vocabulary, particularly when dealing with the affects of music, is the poorer for it.

Last night (early this morning), after writing all of the above, I was still thinking about this (probably helped by the three shots of espresso that I had at a friend’s house after dinner), and came up with a crude way of devising a scale of timbre, in which a tone would be assigned a timbral value between two predetermined extremes for any given voice, so that in our saxophone example, we would precisely describe the timbre of a pure saxophone (think of Paul Desmond) as 0 with a raunchy, gritty R and B style as 12. There are problems with this, as not all timbral variations would fall into these parameters, with some having their own peculiarities that might rate a 2 looking at the partials given in the range, but are clearly more complex, due to some other set of noise elements. At least with a crude system, we would be able to start to analyze the use of timbre in terms of structure, which could yield interesting results. I don’t think that this system would be effective to notate music for reproduction, but would be a good way of analyzing already recorded sounds, such as jazz solos.

In short, I do not have the answer, but it is a topic I have struggled with as an electronic composer, as a theorist, and even as a composer and performer on conventional instruments. Thanks for bringing it up, and I hope that our readers will have something to add.

Posted by erik at May 4, 2003 11:09 PM | TrackBack