Sounds Natural (Part 1)
The Acoustic Guitar
We begin a new series that looks at how to recreate the sound of acoustic instruments using various methods of synth programming and sampling. In the first part, Howard Massey turns his attention to the acoustic guitar.
If you've ever tried to emulate the sound of an acoustic instrument with a synth or a sampler, this new series is for you. In the first of three parts, we focus on the acoustic guitar: what it sounds like, why it sounds that way, and how to duplicate it.
THE ACOUSTIC GUITAR is a stringed instrument which consists of a hollow, hourglass-shaped body (usually made of wood) to which is attached a solid neck. The neck is usually made of wood, and is sometimes reinforced internally with a steel rod (although this is rarely the case with nylon-string guitars). The guitar has six (or occasionally twelve) strings, usually made of steel, bronze, nylon, gut, or combinations of these materials (such as bronze wrapped around a steel core). These stretch across both the body and the neck of the instrument and are attached to a bridge at one end, and pass over a metal, plastic, or ivory nut at the other end, where they are wound around a geared tuning peg. The front panel or face of the body usually has a sizeable sound hole in it to allow the vibrating air within the instrument to escape. In the so-called "f-hole" guitar, there are instead two small f-shaped holes for this purpose.
The neck of the guitar is fretted - that is, small pieces of metal (called, strangely enough, frets) are embedded in the fingerboard. The action of pressing down on a string behind a fret will, of course, alter the vibrating length of the string and thereby change the sound's pitch. If a string's effective length is halved, it will vibrate at twice the frequency, yielding a pitch exactly one octave higher than the sound produced by its vibrating over its full length.
Therefore, 12 frets are placed along one half-length of the string's length, with the 12th fret marking the string's mid-point. Frets 1-11 are situated so as to produce perfect semitone intervals within an octave. If this half-length of the string is halved yet again (resulting in a length equal to one-quarter of the total string length), the pitch will rise yet another octave.
Of course, shorter string lengths require that the fret spacing be closer together, which is why the frets at the top of the neck (the area' closest to the bridge) are more narrowly spaced than those at the bottom. Many acoustic guitar necks, particularly those on steel-string guitars, are long enough to accommodate nearly two octaves of frets.
The strings themselves are tuned in consecutive fourths (that is, five semitones apart), except for the second and third strings (the first string being the highest pitched, and the one closest to the floor). The second and third strings are tuned to a major third (that is, four semitones apart). The actual pitches of the strings are usually - from lowest to highest - E, A, D, G, B, and E (two octaves higher). This tuning pattern may be altered at the player's discretion, yet remains the standard for both classical (nylon- or gut-string) and folk (steel-string) guitars.
In the case of the 12-string guitar (which always has metal strings), the four higher strings are actually the equivalent of two sets of the six-string guitar's two high strings (E and B). Each of these two sets of strings is tuned in unison. The lower eight strings comprise four sets of two each. Each set's pitch is equivalent to that of the six-string guitar's lower four strings (G, D, A and E), but with one string of each set tuned an octave higher than the other.
In all cases, the guitar is tuned by turning the tuning pegs. This has the effect of increasing or decreasing the string's tension, and thereby raising or lowering its pitch.
YOU PROBABLY DON'T need reminding that the acoustic guitar is usually either rested on the knee or attached to a strap worn around the player's shoulder, with the neck normally at the player's left. Guitars are also made for left-handed players, with a neck designed to be held to the right of the player. These instruments feature strings in reversed order, so that the higher strings are still closest to the floor. For the purposes of this discussion, we will refer only to the standard right-handed guitar.
The player changes the vibrating lengths of the various strings by placing the fingers of the left hand on the neck of the instrument just behind the desired frets. This, of course, has the effect of altering the pitch of the guitar sound.
With the right hand, meanwhile, one or more strings may be sounded by plucking or strumming. Either or both of these techniques may be employed at any time, and are executed with the fingertips or fingernails - or with a small plastic or metal device (called a plectrum or, more commonly, a pick), that is held between the thumb and one or two fingers. The greater the force with which strings are strummed or plucked, the greater the amplitude of the resultant vibration, and the louder and brighter the sound.
The particular point on a string at which it is excited has a clear effect on the timbre of the string's sound. The most common playing area is near the sound hole, about a quarter of the way from the bridge to the nut. When excited at this point, the strings yield sounds with the richest and warmest overtone content. When the strings are played nearer to the bridge, however, they tend to generate sounds with more high overtones than low ones - resulting in a brighter sound. Likewise, playing further from the bridge produces less bright tones.
There are also several specialised techniques associated with playing this instrument. The player can physically bend the string being played, thus increasing its tension, and resulting in a rise in pitch. This technique is called note or string bending, and is one that's been emulated by synthesists everywhere. The guitar player can also induce a vibrato effect by vibrating the finger that is placed behind the fret.
A pure harmonic overtone (with no fundamental) can be derived by placing the left finger gently at one of the nodal points of the string (that is, the half-length point, the third-length point, the quarter-length point, and so on) and at the same time, plucking the string with the right hand. Artificial harmonics can also be generated by playing the instrument with a pick held just beyond the fingertip, and simultaneously bringing the finger in brief contact with the string. This action effectively damps the fundamental vibration of the string.
A vibrating string can also be re-excited with the left hand through the use of hammer-ons and pulloffs. These techniques are used to produce legato notes. To perform a hammer-on, the player sounds a string, either fretted or open (unfretted), in the usual manner and then "hammers on" a left-hand finger at a higher fret. This naturally causes the string to sound a higher pitch, but with a gentler rearticulation than if the two pitches were sounded by plucking them independently.
A pulloff is simply a left-hand pluck: a fretted string is sounded in the normal manner, and then the fretting finger is "pulled off' to allow the open string, or a lower fretted note, to sound. Hammer-ons and pulloffs are often used in combination to produce trills and other effects consisting of notes in rapid succession. Strings can also be selectively damped either by decreasing the pressure of the fretting finger (see below) or by resting the heel of the right hand against the strings while plucking or strumming.
THE SOUND OF the acoustic guitar, like that of all plucked or percussive stringed instruments, is generated by the action of a string being moved from its resting point, and then being released. This, of course, results in the generation of vibrations. These vibrations are transmitted from the string (which in itself produces little sound) to the bridge. From there they are translated to the face of the instrument's body, causing the body itself to begin vibrating. These movements are finally transmitted to the air trapped inside the body. The bigger the body, the lower the frequencies it is able to resonate (or vibrate in sympathy with). This explains why a large guitar sounds warmer, or fuller, than a smaller one.
In any event, the vibrating air within the body of the instrument eventually escapes through the sound hole - and combines with the air that has been set in motion by the body's own vibrations.
When this complex movement of air reaches our ears, we are able to perceive the guitar sound. Thus, the body of the acoustic guitar acts as a resonator, while the sound hole gives the sound directionality.
Again, in common with that of most other stringed instruments, the acoustic guitar sound displays a certain amount of inharmonicity. Its predominant overtones, therefore, are usually not simple harmonics, but multiples of the fundamental which are slightly sharp of the true whole-number harmonics. Thus, a fundamental frequency of 110Hz may well yield a second partial of 220.2Hz - not a large discrepancy, but a discrepancy nonetheless. This inharmonicity is more apparent with shorter string lengths (higher notes) than with longer ones (lower notes), and may well vary from string to string, according to the thickness of the string.
As with other wooden instruments, atmospheric conditions such as temperature and humidity affect the timbre of the guitar's sounds. Timbre is also affected by the physical composition and condition of the strings. Steel strings produce more high overtones - and hence brighter sounds - than do nylon strings, while old or dirty strings of any type generate fewer high overtones, hence duller sounds.
Another factor which affects the timbre of the guitar sound is the manner in which its strings are excited. Strings which are plucked with greater force appear not only louder, but brighter as well. Guitar picks made of hard material "give" less than soft picks, and so generate shorter, quicker string excitations - again yielding a brighter sound.
As mentioned above, the point on the string at which it is excited is also a determining factor in the overtone makeup of the sound that's generated. If a string of a well-constructed, nylon-string acoustic guitar is plucked at its usual quarter-length distance, partials 1-10 will be present in the sound. Here, the fundamental is little more than twice as loud as the second partial (which, remember, is just a bit sharper than the true second harmonic due to that inharmonicity). The second partial is little more than twice as loud as the third. Interestingly, both the fourth and eighth partials are virtually absent. The fifth and seventh partials are half as loud as the third, and the sixth partial is just a bit louder than both of these. Here, the ninth and tenth partials have negligible presence.
The steel-string acoustic guitar, as you might suspect, has a very different overtone makeup. First of all, overtones up to the 25th partial are present in the sound in substantial amounts. The second, third, and fourth partials are all present in significant strengths, with the second and third partials are actually stronger than the fundamental in the early stages of the sound. The third partial fades away somewhat more rapidly than the others, but all the lower overtones up to the 13th are substantially present - except for the sixth partial, which remains at a low level throughout the duration of the sound.
The relationship between the fundamental frequency and partials 2-9 remains pretty much the same throughout. Overtones above the ninth partial are gone completely by about half a second into the sound.
Clearly, then, the steel-string acoustic guitar generates a much brighter and harmonically richer sound than the nylon-string guitar.
BEING A STRINGED instrument that is never bowed, the acoustic guitar is non-sustaining. The vibrations of its strings can last for quite a long time, however, if they are not damped by the player's hand. The duration of these vibrations naturally increases in longer strings (that is, lower notes vibrate longer than higher ones). A plucked low note played on an open string on a good nylon-string guitar, for example, can vibrate audibly for a good 20 seconds after its initial excitation.
The attack time of a sound generated by the acoustic guitar is normally very fast, but it may be slowed down slightly by particularly gentle strumming - or when played with a soft plectrum that "gives" with the string.
Decay time can be quite substantial and, once again, will increase for lower pitches. There are actually several factors which affect the decay time. One important factor is whether or not the string is fretted (open strings have much longer decay times, since these do not come into contact with the energy-absorbing surfaces of the neck and the player's finger). Another important factor here is the degree of pressure placed on the string by the finger behind the fret. Decreasing this pressure serves to damp the vibrations, as the string then has greater contact with the soft surface of the finger than with the metal fret. In every instance, however, the overtones of a guitar sound decay long before the fundamental, with the highest overtones diminishing first.
In a typical nylon-string guitar note, for example, most of the higher harmonics fade away within a second. The same general pattern is exhibited by the steel-string guitar, except that the higher overtones linger a bit longer.
Once again, as with all other stringed instrument sounds, here both the fundamental frequency and the amount of inharmonicity of a note are greatest at the note's onset. This is due to the momentary stretching of the taut string as it is plucked or strummed. This causes the pitch to be a little bit sharp and creates increased inharmonic content in this first instant of the sound's existence.
Both of these effects should be taken into consideration when synthesising an acoustic guitar sound, whether it's via digital means or using an analogue subtractive method we'll discuss now.
ASSUMING YOUR ANALOGUE synthesiser has sufficient power, put both of your oscillators to work here, with each contributing a part of the overall sound of a nylon-string acoustic guitar.
Oscillator 1 is used to provide the body of the sound in our subtractive "patch", so set it for a fairly narrow pulse wave (about 10-20%) tuned to its middle register (8'). Oscillator 2 contributes a bit of the initial "pluck" sound, so set it to a sawtooth wave, and tune it three octaves and a major seventh above oscillator 1. This tuning works well here because the seventh is an overtone found naturally in the vibrating string timbre - yet it is far enough from the fundamental to sound "separate" from the main tone. This creates the auditory illusion of a "pluck".
At your synth's mixer section, maximise the volume of the first oscillator while setting the second oscillator, responsible for the "pluck", at about one-fifth volume.
The low-pass filter cutoff frequency should be set at a bit less than half, with no resonance. Route a keyboard controlling signal of about 75% to the filter in order to make higher pitches somewhat brighter than lower pitches. The filter EG serves to fade out the higher overtones rapidly, but the nylon-string guitar is not a particularly bright or timbrally complex sound, so we won't need a great deal of EG depth. Set this at about 2 on a scale of 1-10.
The EG settings are pretty straightforward: set the attack at a bit longer than instantaneous, with a moderate decay, no sustain level, and a release time just a bit longer than the decay time. The amplifier EG should be given the same relative settings, with slightly longer decay and release times. With this kind of envelope setup, you can play legato or staccato notes to create slightly different sounds (the legato notes will fade away more rapidly and the staccato notes will ring out a bit longer).
You'll also want to route an LFO sine or triangle wave through a controller to give you the option of creating a broad vibrato effect at the end of held notes - a characteristic of the acoustic guitar sound. The key here, as with most other imitative patches, is to articulate the sound as a guitarist would - with arpeggiated and strummed chords of no more than six notes at a time.
Finally, if your synth has a velocity-sensitive keyboard, you'll want to route equal amounts of that signal to both filter and amplifier, so that as notes are struck harder, they are both louder and brighter.
IF YOU HAVE access to a digital synth that makes use of Phase Distortion synthesis (ie. a member of the Casio CZ range), this is the patch for you.
To start off with, use the 1 + 2' line configuration. Select the double sine and pulse wave for DCO1, and the square and pulse wave for DC02, to create a rich palette of overtones - one which (because of the double sine wave) also contains a great deal of fundamental frequency. Detune the two DCOs slightly to liven up the sound a bit, and set the DCO envelope to provide a sustained, steady pitch.
The DCW envelopes are each set for a percussive shape with a gradual decay. However, the DCW2 envelope has a slightly slower attack time, a faster decay, and a lower sustain level. Offsetting the envelopes like this helps create a more complex pattern of harmonic change in the total sound. You won't need any keyboard following for either DCW, since you want higher notes to be significantly brighter.
The two DCA envelopes are set quite differently from one another. DCA1 simulates the picking action of the guitarist (actually almost a double-picking effect), while DCA2 produces a short, percussive shape. Apply a bit of keyboard following to both DCAs, since this will shorten these envelope movements slightly as the pitch rises.
This patch should be articulated with a short, strumming pattern to achieve the most realistic sound.
THIS PATCH - FOR users of Yamaha DX synthesisers - simulates the sound of a strummed nylon-string guitar. Use two discrete systems, one for the body of the sound, and the other for the characteristic "thud" heard in the body when the finger makes contact with the string.
The carrier responsible for the body of the sound must generate a complex series of overtones - so a stack and a single modulator are required. The second carrier, which needs only a single modulator, produces the "thud" sound. This configuration is represented in algorithm 8 on the DX7 - so it is naturally the best choice for this particular patch.
Let's first examine the stack which produces the body of the sound (operators 3,4, 5 and 6). Set up a frequency ratio of 3:1 between operators 4 and 3, and use a ratio of 12:3:1 (or 4:1:0.33) for the stack itself to induce high, even-numbered overtones. Detune the top operator to add a slight beating effect to the sound, and add a very small amount of LFO amplitude modulation with a slow-moving sine wave.
Operator 4, the single modulator in this system, has the feedback loop in this algorithm. Set the loop for maximum feedback, but attenuate the output level of that operator slightly to avoid noise and distortion in the final sound.
The EG settings for the operators in this system are all pretty straightforward, with none having any sustain level (L3). All the EGs exhibit a fading pattern, with operator 5 dying away more rapidly than any of the others. This causes the overtones in this sound to fall away more quickly than the lower ones, induced by operator 4. Note that while the nominal output level of the carrier (operator 3) is at maximum, none of its EG levels is at maximum, meaning that its EG movements are somewhat accelerated. The EGs of the carrier and of modulators 4 and 5 should be scaled slightly, so that the higher the note, the more rapid the volume and timbral changes.
Use small amounts of velocity sensitivity in both the carrier and the modulators, so that as keys are struck harder, the sounds become brighter and a bit louder. Apply keyboard level scaling to the modulators, with operators 4 and 5 attenuated slightly from lowest note to highest, and operator 6 (the top operator in the stack) rolled off at both extremes of the keyboard range. Although this patch sounds reasonably good over the full range of the keyboard, it sounds best in the middle registers as a result of these scalings.
The system creating the characteristic "thud" sound is set to a fixed frequency of 117.5Hz. The frequency ratio here should be 1:1, so that all the harmonic overtones will be heard, however briefly, in this extremely transient component of the sound. Add a slight amount of detuning to the carrier, and set the oscillator key sync off, so each key depression produces a slightly different set of overtones. Use a small amount of velocity sensitivity for the carrier, so that harder struck notes will have slightly more "thud".
The EGs in both carrier and modulator are set to yield a transient sound with a short release if notes are played staccato. You'll want to use keyboard level scaling for the modulator to keep this component from sounding overpowering in higher notes, but use no rate scaling here to ensure that the length of this "thud" remains consistent throughout the keyboard range.
Finally, set the LFO for a slow vibrato effect, with a small amount of pitch modulation sensitivity. This should be routed through a controller to allow for selective vibrato effects in your sound.
Articulate this patch accurately, and you've got a pretty convincing nylon-string guitar sound.
THE ACOUSTIC GUITAR is capable of producing very bright transient sounds, so it's always best to mic it with a condenser microphone like the Neumann U87 or AKG 414. Certain dynamic mics, such as the Beyer M60, will also yield good results. Occasionally, an acoustic guitarist may employ a contact pickup, although this is more commonly used for live performances rather than in recording situations. These should rarely be used for the purpose of sampling, as they deliver a very unnatural sound - unless you're used to listening to an acoustic guitar with your ear next to its body.
The microphone should be placed some 2-12 inches from the instrument, near the sound hole, but at an angle in order to prevent a "booming" effect. You might also try positioning the mic closer to the bridge, which will result in a somewhat brighter sound. As always, spend some time finding the optimum mic placement for the type of sample you want to create. Naturally, the distance at which the mic is placed from the instrument has a considerable effect on the sample: greater distances will increase the amount of reflected sound and therefore make the sample more ambient. However, if ambience is what you're really after, it's often better to place the microphone in an omnidirectional or figure-eight configuration (as opposed to cardioid or hypercardioid) and leave it close to the strings.
Typical outboard signal processing techniques include the use of small amounts of EQ to reduce boominess (attenuate the 60-100Hz area) or boxiness (attenuate the 500-900Hz area) and occasionally the addition of brightness (boost the 10kHz area). Small amounts of compression (slopes of 3:1 or 2:1) can be helpful as well, to add "punch" to the strummed or plucked sound. Small amounts of short reverb may also be added to suggest ambience in a dose-miked acoustic guitar sound that lacks any of the natural variety.
Depending on the capacity of your sampler's memory, you should take between two and four samples per octave. (The octave span of an acoustic guitar, incidentally, is equivalent to MIDI note values 40-83.) You'll also need to choose which string you want to sample when changing pitches, as a note on one string can often sound different from the same note played on a string of a different gauge. In general, you should opt for the string on which that note would most often be played in the context of the musical piece you are planning to do. This, of course, may not always be possible, since you may be taking samples for archive purposes, with no particular composition in mind. In that case, your ear will have to be your guide.
Sampling different articulations of the acoustic guitar is usually not necessary, though if your sampler has sufficient memory, you may want to take samples of the same note played with different dynamics or on different strings. The sample length will, as ever, increase for lower notes, and it can be six seconds or more for the lowest notes, making looping necessary if you are working with limited memory. If you need to loop your guitar sample, short loops of just a few wave cycles usually work best, and these should begin well past the peak of the attack (at least a second into the sound). If you truncate after the loop point, you'll need to use your sampler's filter and amplifier envelopes to restore some of the natural decay and release portion of the looped sound. Often, judicious use of the filter envelope to decrease the upper harmonic content of the sample is helpful as well.
Next month, we'll be adopting the same approach (discussing the nature of the acoustic sound, programming three types of synth, and sampling) for another often-emulated instrument, the clarinet.
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.
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!