Sounds Natural (Part 2)
The second instalment of our series on programming and sampling the sounds of acoustic instruments. This month Howard Massey turns his attention to Acker Bilk's favourite: the clarinet.
In the second part of our series on emulating the sound of acoustic instruments, we take a close look at the clarinet and find out how it relates to synthesisers and samplers. Text by Howard Massey with Alex Noyes and Daniel Shklair.
THE CLARINET IS a reed instrument. Its body is made from wood or plastic and consists of tube around twenty-five inches long, tapered at one end and attached to a mouthpiece. This tapered end makes the clarinet unique among woodwind instruments in that it is a cylindrical bore "closed tube" instrument (although it is actually only partly closed). The other end is connected to an open and flared segment which forms the clarinet's bell. The mouthpiece, as its name implies, is held in the player's mouth where air is introduced into the instrument. A single cane reed is attached to the mouthpiece with a device called a ligature (a light metal ring which fits over the mouthpiece and is held in place with screws). Because there is only one reed, the clarinet is classified as a single-reed instrument (as opposed to a double-reed instrument like the oboe).
The clarinet's body has a series of holes which can be covered and uncovered by the player's fingers or by the action of padded keys. The opening and closing of these holes changes the length of the column of air inside the instrument, thus changing the pitch of the sound produced by the clarinet.
The clarinet family comprises the clarinet in E-flat (soprano clarinet), clarinet in A, alto clarinet in E-flat, bass clarinet in B-flat, and contrabass clarinet in B-flat (pedal clarinet), all of which are monophonic instruments. The note range of the clarinet is D3 to F6 (MIDI note values 62-101).
ALL REED INSTRUMENTS use the same method to produce sound: A reed (usually made of cane) is made to vibrate by blowing air across or into it, and the resulting vibrations are transferred from the mouthpiece to the column of air in the adjoining tube. This process is much like the one that occurs when air is blown past a blade of grass held between the two thumbs, causing the blade itself to vibrate and produce a high-pitched squeaking sound. Specifically with the clarinet, the player puts the mouthpiece and reed into his or her mouth and presses the lower lip against the reed in order to partially close the reed opening. The player then blows air into the mouthpiece and, when the level of air pressure is sufficient, the reed will begin to vibrate and consequently produce a tone. If the amount of air flow falls below a certain level, the sound will stop. The air flow is started and stopped by the clarinetist with a technique called tonguing, or by changing the lip pressure on the reed. This regulation of air flow, if performed periodically, will also create vibrato effects in the clarinet's sound.
The clarinetist's embouchure (the way that the reed and mouthpiece are held in the mouth) and the amount of lip pressure on the reed can, along with the thickness and/or stiffness of the reed, affect the timbre of the sound produced. Embouchure techniques also help the clarinetist to create subtle pitch changes and vibrato, although the pitch of the instrument is more often varied by covering and uncovering the holes in the instrument's body. The loudness and overtone content of the clarinet sound can be increased by blowing harder across the reed, and, as with all wind instruments, the louder the sound, the brighter the timbre.
On the whole, the clarinet is one of the more facile and responsive wind instruments, allowing for a wide range of sophisticated articulations. It is capable of dramatic pitch-bends - induced by adjustments of embouchure. Another speciality is the ability to perform rapid glissandos (accomplished by nimble fingering techniques).
IN ALL WIND instruments, the fundamental frequency of the standing waves generated is determined by the length of the tube. Because the clarinet is a closed tube instrument, however, the frequencies it produces will be one octave below that indicated by its length (that is, an octave lower than the flute, for example, which is an open tube of similar length).
The clarinet exhibits no particular emphasised frequency area (unlike, for example, its cousin, the oboe) and no concentration of energy at any particular harmonic. As such, the timbre of the sounds it produces is a relatively simple one, containing all the harmonic overtones, but with the odd-numbered harmonics at much higher amplitudes. The waveshape that is thereby generated is similar to the classic square shape which is easily emulated by both analogue and digital oscillators (see waveshape illustration below).
In our timbral analysis of a clarinet playing an F3 at a moderate volume, the fundamental is predominant throughout, with the third, fifth, ninth, eleventh, and thirteenth harmonics present in substantial strengths. Small amounts of upper harmonics (up to nineteen times the fundamental frequency) are also detectable in lesser strengths. Of particular interest is the surprising strength of the tenth harmonic in the early stages of the sound (although this pretty much disappears by the time three-quarters of a second has elapsed) and the uniform weakness of the seventh harmonic.
As the clarinet produces louder sounds, higher harmonics are generated, and in greater strengths (relative to the fundamental). As the pitch of the sound rises, however, an unusual thing happens: the strength of the fundamental relative to these higher harmonics actually increases. In other words, low pitches played at moderate amplitudes will be quite rich in overtones, while high pitches played at a lower volume will be considerably purer and "thin" sounding because they contain relatively more fundamental frequency. In fact, very soft notes played in any register have virtually no harmonic content. The clarinetist can thus produce a wide range of timbral variation. This ability, along with the advantages of the several different playing techniques, makes the clarinet one of the more expressive wind instruments.
THE CLARINET SOUND can vary considerably from player to player - and from one note to the next. In general, however, the louder the note, the faster the initial attack time (although a player can also build a note's intensity during the sustain portion of the sound). The clarinet, like all wind instruments, is a sustaining instrument, meaning that as long as vibrating air is introduced into it, a tone will continue.
As with virtually every acoustic instrument, the higher harmonics generated by the clarinet sound disappear more rapidly than lower ones. Like all other wind instruments, the sound begins with a very brief flurry of enharmonic overtones (a slight "chiff" sound, like that found in the flute) as the standing waves begin their existence and build to a sustaining level.
Certain specialized playing techniques can change the articulation (and therefore, the envelope) of a clarinet note drastically. In particular, a technique called slap tonguing is commonly used. This articulation is accomplished by "popping" the tongue against the roof of the mouth (with the reed in between). This technique changes both the overtone content and envelope of the sound, making it almost percussive in nature and quite rich in the higher overtones.
THE CLARINET SOUND is often likened to that produced by a pure, slightly filtered analogue square wave, since both are made up predominantly of odd-numbered overtones. For this patch, however, we found that the best results were achieved by using two pulse waves, each having slightly different pulse widths. Begin by tuning both your oscillators in unison in their middle register (8') - Set oscillator 1 to a 30% (or 70%) pulse width, and oscillator 2 to a 20% (or 80%) pulse width. Begin by sending all of the first oscillator's signal through the mixer, then slowly blend in oscillator 2 - listening as you do so. At a certain point (about 70% amplitude on the Sequential Pro One), the timbre becomes remarkably clarinet-like. Stop when you reach this point and move on to adjust the filter.
Set the cutoff frequency of your lowpass filter at about the midpoint, and set the resonance at about 25%. Unlike the tone of a brass instrument, there is little timbral change during the duration of the clarinet sound, so you won't be needing the filter EG at all for this patch. Set its depth control to 0. Route about half of your keyboard controlling signal to the cutoff frequency so that higher notes are somewhat brighter than lower ones.
The amplifier EG will have very simple settings. Set it for a moderately fast attack time (but not instantaneous - remember that wind instruments take a little time to "speak"). Set both the decay time and sustain level at about half, and use a short, though not instantaneous, release time.
A little vibrato from your LFO will help to add some realism to the sound, so route just a very small amount of a sine or triangle wave from your LFO via one of your real-time controllers to the frequency inputs of one or both of your oscillators. Articulate correctly (monophonically, with judicious pitch-bending), and you've got an instant "clarinet" at your fingertips!
THE 1+2' LINE configuration is best for this patch because it gives you access to a broad palette of harmonic overtones (even though no detuning is used in the generation of this solo instrument patch). Since the clarinet sound contains an abundance of odd-numbered overtones, you'll want to use square waves for both DCOs, adding a pulse wave in DCO1 and a double sine in DCO2 - for a little bite.
A very rapid drop in pitch is instituted by both DCO envelopes, with slightly different movements in each. This helps to simulate the pitch change during the initial part of the clarinet sound. The DCW envelopes both have simple ADSR-type shapes, each with moderately high sustain levels, so that held notes are fairly bright. A large amount of keyboard following is employed for line 1, and a moderate amount for line 2 to account for the fact that the clarinet sound does not get significantly brighter (and is certainly harmonically simpler) in its higher registers. Adding keyboard following in these differing amounts will make for less obvious timbral variations in the high and low registers.
The DCA envelopes have even simpler shapes, with moderate attacks, fairly high sustain levels, and moderate release times. Small amounts of keyboard following are employed here as well so that higher notes undergo these changes a bit more rapidly than lower ones.
No modulation is used in this patch, and the LFO supplies a subtle delayed vibrato (which will only affect held notes). The entire patch is transposed up one full octave to emulate the customary clarinet register. As ever, articulate correctly (monophonically, with judicious pitch-bending) for the most realistic results.
ALTHOUGH THE CLARINET produces a timbrally simple sound, it is best to use two carriers and two stacks to simulate it accurately. One system serves to emulate the "chiff" that is present at the onset of the clarinet sound (here, this is accomplished with a feedback loop and the appropriate EG settings). Because the typical clarinet sound is surprisingly bright, though hollow, a second stack provides sufficient overtone content to emulate this effect. Algorithm 3 is best suited for these purposes (although algorithm 4 will work just as well, since little feedback is required).
The stack consisting of operators 1, 2, and 3 provides the body of the clarinet sound. Give this stack a frequency ratio of 6:4:1. This arrangement provides you with a broad range of harmonics with large gaps, and large amounts of upper odd-numbered harmonics. Employ a very slow detuning between the carrier (operator 1) and its nearest modulator (operator 2). Assign a small amount of velocity sensitivity to the carrier to allow for performance control over dynamics. Assign a moderately large amount of velocity sensitivity to modulator 2, so that softly struck keys will generate tones with little harmonic content (in imitation of the relatively broad timbral range offered by the clarinet itself). Attenuate the output levels of operators 2 and 3 slightly over the middle and upper registers of the keyboard with the keyboard level scaling control, so that high notes are not overly bright.
The carrier EG is set for a subtle double attack so that held notes build slightly in volume, while the modulator EGs have simple ADSR-type configurations with no sustain levels. Note that modulator 2 has an extremely slow release so that no timbral change is detected after the key is released. Small amounts of rate-scaling are applied to operators 1 and 2, and a large amount is applied to the top modulator (operator 3) so that higher pitches undergo changes in timbre and amplitude somewhat more quickly than lower notes.
The stack of operators 4, 5, and 6, as noted above, contributes mainly the slight "chiff" effect at the beginning of the sound. (Isn't it great to have six operators at your disposal?) Select a frequency ratio of 2:2:1 with a moderate output level for the bottom modulator (operator 5) so the sound is not overly bright. Also assign a small amount of feedback to the feedback operator (operator 6). Clarinet notes played at low volumes will have less of this "chiff", so assign a great deal of velocity sensitivity to the carrier. The EGs, as you might suspect, are all simple percussive configurations, providing slow release times for the modulators so that no timbral change is detected after the key is released. A large amount of rate-scaling is applied to the carrier, and smaller amounts to the modulators, so that the "chiff' sound is shorter at the onset of higher notes. This is done because higher clarinet notes take less time to "speak" than lower ones (this is also true of all other wind instruments).
The pitch EG is inactive in this patch, but the LFO is providing a slow delayed vibrato, which only appears in held notes - another characteristic of the typical clarinet sound. You might also try routing this signal through a real-time controller to achieve the same effect selectively.
BECAUSE THE CLARINET has a fairly broad timbral range, opinions differ as to the best type of mic to use when recording or sampling it. If you're looking for a warm, rich clarinet sound, a good-quality dynamic microphone, or even a ribbon microphone, should work well - but if you want a slightly brighter, more cutting sound, a condenser mic will probably do a better job. When in doubt, you can always turn to either of the two studio workhorses - the Neumann U87 or the Shure SM57. Set them both up, about a foot-and-a-half away from the instrument. Position the mics about eight inches higher than the flared end of the instrument. Now listen and your ears will tell you which mic is giving you the kind of sound you had in mind.
The somewhat unusual sound radiation patterns of the clarinet can pose some unusual problems when it comes to mic placement. Low clarinet notes produce sounds that radiate mainly from the open holes near the flared end - whereas higher notes tend to make the entire instrument vibrate, thus making the sound far less directional (in direct contrast to the way brass instruments, for example, radiate sound). If you place the mic as suggested above, be sure your placement is quite accurate to ensure a fairly uniform output from note to note. This setup will impart a fair amount of room ambience to the recorded sound, which you may or may not want in your sample. If you'd rather close-mic, and not have to hassle with getting the optimum placement for all notes, you can simply move the mic around at a closer distance (while monitoring a single note over headphones) and simply find the best spot for that particular note. If you follow this course of action, be prepared to move the microphone after every note.
If your mic selection and placement are good, you won't need to use much outboard signal processing when recording (or sampling) this instrument. Small amounts of EQ may be added to brighten or warm the sound, but you probably won't need any compression or limiting. If your sample is miced closely and you want a bit of room ambience, you can simulate these acoustics by adding a small amount of reverberation.
As pointed out earlier, the clarinet has fairly distinct timbral variations from register to register - so you'll want to take between two and four samples for each octave, if your sampler's memory permits. You can conserve memory by looping this sustaining sound. Start your loop well after the attack portion (which may be quite long in the clarinet, so listen carefully). A good rule of thumb is to start your loop no sooner than a second-and-a-half into the sound. Once you're into the stable, sustained portion of the sound, short loop lengths should work just fine, and of course, you should truncate the loop after it ends.
The clarinet sound, like that of the flute and oboe, typically contains vibrato during sustained notes. This means that you must decide if you want to sample a sound containing vibrato. If you do sample a natural vibrato, remember that subsequently changing the pitch of the sample will also change the speed of the vibrato. This will sound unnatural beyond a change of a semitone or so, meaning that you will need more samples per octave. Bear in mind also that a sampled sound with natural vibrato will be considerably more difficult (though by no means impossible) to loop. If memory in your instrument is a restricting factor, you'll probably be better off sampling a sound with no vibrato and then using your sampler's LFO to create an effective though predictable vibrato in the sound.
While you're at it, you might consider sampling several different volumes and attacks as well as different playing articulations (like pitch-bends and slap tonguings). You can then use the velocity switches and crossfade controls in your sampler to allow you to use these same performance techniques in your sample.
The Sounds Natural series comprises excerpts from A Synthesist's Guide to Acoustic Instruments, a new book by Howard Massey, Alex Noyes and Daniel Shklair.
More from Music Sales, (Contact Details)
Feature by Howard Massey
Previous article in this issue:
Next article in this issue:
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!