Sounds Natural (Part 3)
In the third and final part of our series on synthesising and sampling the sounds of acoustic instruments Howard Massey hits it off with the snare drum.
In the third and final instalment of our series on synthesising and sampling the sounds of acoustic instruments, we examine the snare drum. Text by Howard Massey with Alex Noyes and Daniel Shklar.
A SNARE DRUM IS a member of the same percussive family as the tom tom, timbale and tenor drum. Essentially, it consists of a hollow cylinder with two membranes attached. The cylinder of the drum (called the shell) is quite shallow, and constructed of metal, wood, or plastic. The membranes stretch across both ends of the cylinder and are referred to as the heads or skins of the drum, as they were originally made of animal skin, though plastic has now taken its place. The heads are held in place by a series of metal lugs, which may be tightened or loosened to alter the tension of the head. Tighter head tensions produce a sharp, snappy, sound (typically used in orchestral or jazz settings). Looser head tensions result in a thick and full snare drum sound (often used in rock settings).
Stretched across the bottom head are a series of strands of thick wire or gut (with up to twenty-four individual strands). These wires are called the snares, and they rattle when the bottom head starts vibrating as a result of the top head being struck. This serves to add the characteristic bright "crack" to the sound of the instrument. These snares can also be disengaged from the bottom head with the push of a lever. When they are disengaged, the snare drum sounds very much like a tom tom.
The diameter of snare drum heads ranges from 16" to 20". The depth of the drum's cylinder may be as little as three inches (in the piccolo snare drum). A cylinder depth of six or seven inches is found in the orchestral snare drum, which is also commonly used in rock settings. The cylinder depth may be as great as eighteen to thirty-six inches in the case of the military snare drum. In general, snare drums of greater depth will produce deeper sounds.
When the top head of the snare drum is struck, the resultant vibrations are transmitted to the air trapped inside. This, in turn, causes the lower head to vibrate sympathetically, which then causes the snares themselves to vibrate. One of the characteristics of the snare drum is its tremendous dynamic range. It is capable of producing sounds ranging from barely audible to nearly as loud as the report of a small explosive.
THE SNARE DRUM is played by striking its top head with a drumstick (usually made of wood), or by hitting or stroking the top head with a wire brush. Some drumsticks are fitted with plastic tips, although these have a much greater effect on cymbals than they do on the snare drum.
The top head may be struck near its centre (but not usually at dead centre), resulting in a full sound. It may also be struck at its edges at the same time as the metal rim is struck. This is called a rim shot, and produces a sharp "crack". Obviously, when the drumstick is brought into contact with the head with greater force, the resulting sound will be louder. Heavier sticks, or even the blunt back ends of drumsticks, are often used to impart even more force to the blow.
There are also many different strokes used by drummers. The head may be struck by a single drumstick, or it may be struck nearly simultaneously by two sticks (this is called a flam). It may be struck repeatedly with alternating sticks (a single-stroke roll), or repeatedly with two strikes of one drumstick, followed by two strikes of the other (a double-stroke roll). The former technique is often used in contemporary rock settings, and the latter more for military and orchestral playing. Drummers may also occasionally place the tip of the drum stick near the center of the drum and drop the blunt end of the stick onto the rim (called a sidestick). This produces a sound very much like that of a resonant woodblock. There are many more types of drum strokes commonly used as well.
In rock settings, the snare drum is the predominant drum. Here, it provides the backbeat - meaning that the drum is regularly struck on the second and fourth beats of each four-beat measure. In jazz settings, the snare drum is more commonly played "around" the beat, with one stick or brush often lightly filling in with syncopations. In orchestral music, the snare drum is used more for specific accents, and to provide dynamics to the music. The instrument is a staple of military music, where its characteristic rolls and flourishes add much to the flavour of the music.
THE SNARE DRUM sound has no definite pitch. Instead, it contains a broad range of enharmonic overtones ranging in frequency from about 100Hz all the way up to 20kHz and beyond. The particular metal shell snare drum sound that we sampled and examined, exhibited inharmonic overtones up to approximately 121 times the 100 Hz fundamental frequency. The overtone content of a specific snare drum depends upon many factors. These include the construction of the shell, the thickness of the top head, the force with which it has been struck, and the location at which it was struck - with the last of these having perhaps the greatest effect on the overall timbre. Striking the head at dead center yields the dullest sound, with a predominant fundamental and slightly subdued overtones. Striking the head just off centre seems to lessen the strength of the fundamental just a bit, while increasing the amplitude of the higher overtones. As the snare drum is struck closer to its edge, the fundamental is attenuated further still, while the higher overtones increase in strength, resulting in a thin and bright sound.
Although there's little discernable pitch in this sound, the fundamental frequency, as noted above, can be altered by changing the tension of the head. This is accomplished by tightening or loosening the tuning lugs which hold the rim tightly over the head and against the drum body. There are generally eight to twelve of these lugs, and it is usually desirable to apply equal tension to each in order to produce the most pleasing tone from the instrument. As the head is stretched tighter, the fundamental frequency rises and as the tension of the head is relaxed by loosening the lugs, the fundamental frequency will fall. The thickness of the material used for the head has considerable impact upon the timbre of the sound. Thicker skins generate fewer high overtones than thinner ones (since they absorb the vibrational energy more readily). These and various other anomalies make it difficult to generalise about the timbral makeup of the typical snare drum sound, but certain features are common to most.
The overtone structure of any vibrating circular membrane is quite complex. The nodes of vibration travel across the surface of the head in different circular and linear configurations. The circular patterns are concentric, while the linear patterns often cross one another to produce enharmonic, rather than harmonic, overtones. Some of the overtones produced by simpler vibrational nodes occur at 1.59, 2.13, 2.29, 2.65, and 2.91 times the frequency of the fundamental. These non-integer components, and many more, are present in great strength, making the snare drum sound closer to pure, unpitched white noise than perhaps any other acoustic instrument.
THE SNARE DRUM sound is of brief duration (generally lasting less than a second) - a non-sustaining and transient sound. The drum's head is stretched when struck by the stick and snaps back almost instantly - producing no more sound until it is struck again. Because of its brevity, this sound produces virtually no periodic wave.
The sound's attack and decay times are virtually instantaneous, with no sustain. The snares have a damping effect on the bottom head, causing a very short release time. The snares themselves also quickly cease vibrating. This basic shape characterises the timbre, as well as the loudness, of the sound, although the higher overtones of the snare drum sound are the last components to fade away.
BECAUSE THE SNARE drum sound is almost entirely without pitch, you need no audible oscillators for this patch, so shut off their signal completely at the mixer. Then, raise the output of your noise generator to maximum. This will create a random, unpitched timbre.
Set the low pass filter cutoff frequency at about 30% (but adjust this by ear for your particular subtractive synthesiser). Add no resonance, and no keyboard control, so you can generate a fixed timbre throughout the keyboard range.
Since the snare is such a bright sound, it may surprise you to see such a low filter cutoff setting. However, the filter EG compensates for this setting by sweeping the frequency considerably. Set this for maximum depth, an instantaneous attack, a decay time of about half of its maximum, no sustain, and a very short release. You should adjust the decay time further by ear, as it has a great effect on the type of snare drum sound produced with this patch. Set the amplifier EG with very similar values, but make sure its decay time is a bit shorter than that of the filter, as you don't want to be able to hear the filter close down completely at the sound's end. These EG settings will allow you some articulation, as playing staccato notes will result in short, transient sounds.
Further articulation controls are available to you if you are working with a velocity-sensitive keyboard. If so, route some of this controlling signal to the amplifier and/or filter. This will cause louder and/or brighter sounds to be produced as keys are struck more quickly. If your system has a high pass filter, you may find that removing some of the lower overtones in the noise may help the "crack" of the snare drum sound - a result of the fundamental dying away rapidly, leaving only the higher overtones at the sound's end. You might also want to try adding a very small amount of one of your audio oscillators, tuned to a low, audible range frequency (90 to 100Hz), and using a triangle wave. This should help to add some "thud" to the "crack" being contributed by the noise generator.
Finally, if you have access to some outboard signal processing equipment, you will undoubtedly find that some reverb or short feedback delay helps this patch a great deal.
THIS RELATIVELY STRAIGHTFORWARD patch uses the 1+2' line configuration with two harmonically rich waveshapes (the sawtooth and pulse). A great deal of detuning should be employed here, as well as noise modulation for the contribution of the snares themselves.
Both DCO envelopes are set to produce a sharp drop in pitch throughout the brief duration of this sound, in emulation of the typical pitch change which occurs in the sound. Set the DCW envelopes to very simple percussive shapes, with no keyboard follow for line 1, and only the minimum amount for line 2. This renders higher pitches considerably brighter than low ones - allowing you to synthesise various types of snare drum sounds in the different registers of the keyboard. You can then pick the one note that best represents the sound you want. We found this patch to be best between the lowest and second lowest octaves of the keyboard, but your ears may tell you differently.
Give the DCA envelopes percussive settings as well, to cause the sound to exist for only a very brief period of time, in imitation of the transient nature of the snare drum sound itself. Again, use no keyboard following for line 1, and only a very small amount for line 2. These offsets mean that part of the sound will undergo more rapid changes for higher notes - while another part of the sound changes consistently throughout the keyboard range - helping to add complexity (and therefore realism) to the final sound.
You do not need to use the LFO here a all. However, if you play this patch on a velocity-sensitive synthesiser (like the Casio CZ1), route some of this controlling signal to both the DCA and the DCW. This will give you a great deal of dynamic control - important with an instrument with as great a dynamic range as the snare drum.
VERSATILE AS THE digital FM system is, it's not particularly good for generating random timbres or noise effects, because the system becomes erratic and unstable when distorting a sound to this extent. Creating this patch was therefore perhaps the greatest challenge of all of those presented in this book. The resulting sound is a pretty realistic imitation of the sound of a military snare drum.
As with the subtractive and digital phase distortion patches, white noise forms an important basis for this sound. This is because its lack of pitch and its strength in high frequency components make it the ideal sound source from which to generate the sound of the vibrating snares. One good way of generating white noise in an FM system is to use a stack of three modulators to feed a great deal of signal into a single carrier. This results in an overloaded and distorted timbre. With the luxury of the six operators available in the DX7, two operators can then be allocated to produce the hollow "drum" part of the sound. Algorithm 1 provides the necessary configuration. We chose this specifically over algorithm 2 (which is virtually the same) because it provides a necessary feedback loop on the stack - rather than on the single modulator-carrier system, as does algorithm 2.
All operators are set in a fixed frequency mode, resulting in a single snare drum sound throughout the keyboard range. This allows you to play this sound percussively. Let's start by examining the simpler system - consisting of operators 1 and 2. A snare drum's fundamental frequency is approximately 100Hz, so set operator 1's fixed frequency in that area. You can adjust the tuning of this patch by raising or lowering this frequency slightly, but we got the best results at 91.2Hz. Set Op 2 about 30 cycles per second lower, to generate a wide variety of inharmonic undertones, as well as overtones. Adjust its output to a fairly low level to produce a warm and hollow timbre. Apply no velocity sensitivity to either operator. Since you need a consistent sound throughout the keyboard range, there is no need to use keyboard level scaling controls either. The EGs of both of these operators should be set to a fairly simple percussive shape, with Op 2's EG being slightly more contoured. Both EGs have a short after-ring, with that of the modulator relatively longer than that of our carrier - so you never quite hear the end of this timbral shift. Again, because you are creating a consistent sound throughout the keyboard range, use no keyboard rate-scaling here either.
So much for the hollow "drum" component of the sound, now let's see how to make the sound of the snares with the other operators. As mentioned earlier, this process requires the use of three stacked modulators feeding a great deal of signal into a single carrier. The idea here is to eliminate the fundamental frequency being produced by the carrier - which, in theory, is not truly possible. However, this effect can be achieved by placing the fundamental frequency of the carrier in the subaudio range (10Hz, in this instance). Then, do the same for the two modulators immediately above it. Set the top modulator in the stack, which has its feedback loop wide open, to an audible range frequency (707.9Hz, in this case) to produce something more than a rumble. All of these modulators are placed at or near their maximum output level to induce a great deal of distorted sound. With these settings, something close to white noise is produced.
Use the envelope generators in these four operators to shape this white noise into something approaching the sound of the rattling snare. This is easily accomplished by setting the carrier for a basic percussive shape (with a small amount of rate 4 after-ring). Retain the basic square EG shapes in both Ops 4 and 6 - and add just a bit of different after-ring value to each. Then, adjust the EG of the middle modulator (Op 5) to a percussive shape similar to that of the carrier. As before, use no rate-scaling or keyboard level scaling, but add some slight velocity sensitivity in the carrier (Op 3), so that as keys are struck harder, notes will have a bit more "snare" sound.
The last important component of the snare drum sound is the characteristic pitch shift found in all vibrating membrane instrument sounds. The typical digital FM system (like that of the DX7) provides you with a separate pitch EG. The problem is that the pitch EG does not affect any operators in fixed frequency mode - and all the operators in this patch are in that mode. The LFO provides a solution. Set it for a sawtooth-down waveshape. The beginning part of this wave is almost instantaneous - this is followed by a slow drop. Slow the LFO speed down to minimum and - most importantly - synchronise it so that each new key depression starts a new wave. Then, use this signal for direct pitch modulation. (This affects all operators equally). This causes the pitch to drop somewhat after each key strike - just as a real snare drum's pitch drops immediately following each strike of the drumstick.
EFFECTIVELY RECORDING THE snare drum is a demanding job for the recording engineer. The task is a complex one, often requiring a great deal of signal processing and more than one mic.
Current fashion favours an ambient snare drum sound. This is usually obtained by placing mics a distance of ten feet or more from the instrument. These should face a reflective surface (a wooden ceiling or floor, or a glass window). The resultant ambient sound is usually combined with a sound obtained by close-miking - facing mics at a distance of a few inches toward the top head, bottom head, or both. Miking the top head will add more "thud" to the recorded sound. Miking the bottom head will result in more "crack", as the rattling snares themselves are here more prominent. Occasionally, undesirable head "ringing" may occur - especially with drums that are not correctly tuned. This can usually be eliminated by adjusting the tuning lugs or by placing a strip of thick tape (typically gaffa tape) along the length of the top head. If multiple microphones are used, be sure that they are all in phase with one another, or serious frequency cancellations may result. Close mics are generally placed at a distance ranging from one to six inches from the drum's head - about a quarter of the distance from the rim to the centre. If you change the mic placement even slightly, the sound will often undergo a considerable change.
The choice of mic is also very important, the most critical factor being whether it is capable of handling the very loud dynamics that the snare is capable of generating. Dynamic mics, such as the Sennheiser 421 and the Shure SM57, are perhaps most commonly used. Condenser mics such as the AKG 414 or the Neumann U87 are also often used to record this instrument, but they must have their attenuating pads switched in. Sometimes the best results are obtained by placing a dynamic mic over the top head and a condenser mic under the bottom head (as the condenser is typically better able to reproduce the very high frequencies that emanate from the snares).
Typical signal processing used includes good doses of EQ, often boosting the 100-200Hz and 12-15kHz areas. This enhances both the "thud" and "crack". Some midrange frequencies may be rolled off as well. Compression is also often applied to give "punch" to the sound. To this end, slopes of 3:1, 4:1, and greater are used, with peak attenuations of anywhere from a few decibels to 10 or more decibels. If ambient mics are used, they are often gated and keyed from one of the close mics - and they may be even more severely compressed, with relatively slow compressor attack times. This adds a "sucking" effect to the sound (see Collins, P. and Bonham, J.). If your snare drum is first recorded on tape, careful tape saturation techniques can add a smoothing effect to the sound which is sometimes preferable to that added by a compressor. In other words, record the sound at as high a level as possible, without causing undesired distortion. Short reverberations and/or slap echoes (echoes with feedback) are usually added as well - sometimes with a slightly delayed send (predelay) so as to separate the reverb from the main signal. Other, more exotic recording techniques include using the snare to key a gated white noise source (typically from an analogue synthesiser) - or to trigger sampled "whip-crack" sounds.
Of course, you have the option of sampling an unprocessed snare drum sound, and processing the sample itself. However, it is more common to sample a processed sound, since this calls for little manipulation of the sound after it leaves the sampler. Looping is unnecessary here, since the snare drum is a non-sustaining and transient sound. Because the pitch area of the snare drum sound does not change much from note to note, multisampling for pitch change is also unnecessary. In general, you should do most, if not all, of the signal processing before the snare sound reaches your sampler. Again, record the sample at as high a level as possible without overloading (be particularly careful here as digital overload is not a nice sound). Take the time to experiment - some of the finest professional recording engineers spend their entire careers searching for the "perfect" snare drum sound.
The Sounds Natural series comprised excerpts from A Synthesises 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
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!