Académique Documents
Professionnel Documents
Culture Documents
Rolf BAder
Institute of Musicology, Neue Rabenstr. 13, 20354 Hamburg, Germany e-mail: R_Bader@t-online.de,
Noninearities in the sound production of the classical guitar are discussed. Based on a physical model
of the whole classical guitar with a Finite-Difference Method, three kinds of nonlinearity appear to be
crucial for a realistic guitar sound. First, the coupling of transversal and longitudinal vibrations in the
guitar body determine the transient role of the different guitar parts (top plate, back plate, ribs ect.) during
the initial transient and the steady-state phase and also determine the brightness of the sound in terms of
internal tensions in the guitar geometry. Secondly, the complicated guitar geometry causes the inclosed
air to separate the income of the different body parts during the initial transient phase to make the sound
much deeper or heavier. And thirdly, the coupling of the strings to the to plate implies nonlinearities
as an essential factor to produce realistic amplitudes of the higher harmonics of the guitar sound. Sound
examples of the modeled guitar show differences in the fine structure of the sounds determining the quality
of the instrument as well as its characteristics within its genre.
1 Introduction
The classical guitar can be described as a string acting on
a resonance body. The string acts on the top plate of the
guitar, which is coupled to the ribs, the inclosed air and
the guitar neck. The back plate is driven by the ribs and
the inclosed air and each part acts back on the top plate
and on the string again. The guitar tone, like nearly every musical instrument tone consists of an initial transient
phase followed by a so called quasi-steady state. The initial transient phase is important not only for the instrument identification by listeners, but also in terms of the
musical expression the player is in. Within the guitar, this
expression consists of a tone loudness, the way, the finger
slides over the string before release to produce a kind of
pre-scratch noise and the amount of overtones in the following guitar sound controlled on the one hand side by
the plucking point and on the other hand side by the curvature of the string at the plucking moment controlled by
the amount of finger nails used by the player.
The initial transient phase consists mostly of a kind of
knocking sound. This is produced by the first impulse
of the string acting on the guitar top plate. This impulse
drives all eigenfrequencies of the resonance body. These
driven eigenfrequencies may interact with the following
eigenfrequencies of the string and cause large amplitude
fluctuations within the initial phase of up to about 50 ms.
So the resonance frequencies of the guitar body, their exact values and their ability to be driven by the first impulse determine much of the character of the initial transient and so of the guitar character as a whole. Also,
because of the initial transient being mostly defined by
the struggling between the resonance frequencies of the
guitar body and the eigenfrequencies of the string, he
exact behaviour of the incoming inharmonic frequencies
and the amplitude shapes of the string frequencies in the
initial transient for one defined played note cannot be influenced by the player and is nearly absolute the same for
each played pitch over all reasonable string amplitude.
The sound of the guitar then is determined by the different guitar parts radiating with different spectra coming in
at different times [1]. The top plate is first and reacts immediately to the string. Its sound has a clear attack but
is judged more like a string simply attached to a wooden
plate and lacks the fine structure in the classical guitar
sound. This fine structure is produced by the back plate
sounding mainly in the lower middle frequency range and
the ribs sounding in the upper middle frequency range.
The air inclosed adds a kind of kick at the lower frequency range. Finally the neck still adds higher frequencies. Also the incoming time of the different guitar body
parts are different. The top plate get in immediately as
mentioned above. The back plate and the ribs only come
in after some milliseconds and the inclosed air linearly increases in amplitude within the first few milliseconds. All
these parts make the guitar tone much deeper and more
profound, suitable to a solo playing style within concerts
halls for what this instrument is mainly thought.
The reasons for most of these fine structures of the sound
are found in nonlinear behaviour of the guitar body or
vibrating parts, where three main findings are presented
here.
Rolf Bader
and is
2 uz
E K2
=
2
t
4 uz
4 uz
+
4
(x
(y 4
(1)
2 uz
E K2
=
t2
a)
!
4 uz
4 uz
+
.
2
2u
(x + xu2x )4
(y + y2y )4
(2)
b)
c)
d)
Figure 1: Comparison between the knocking sound of a
guitar back plate with and without tension. No tension:
a) Development of the spectral centroid of the back plate
without tension of the time series in b). Tension: c) Development of the spectral centroid of the back plate with
tension of the time series in d). The centroid of the tension case is much larger at the beginning of the sound
leading to a brighter and more brilliant attack. The high
value of the case of no tension at about 50ms cannot be
heard very good, because the sound is nearly silent then.
Rolf Bader
ergy is getting into the inclosed air volume. When watching the inclosed air moving inside the guitar within the
initial transient phase by the modelling software visualization screen, it can be seen, that the air takes four cycles to reach the back plate and the ribs. So suddenly at
the fourth cycle, the back plate and the ribs are driven in
their normal vector or radiation direction by the low resonance frequencies of the inclosed air and so start radiating
driven by low frequencies and filtering these frequencies
by their own resonance spectrum again in the upper middle frequency range by the ribs and the lower middle frequency range by the back plate. During this whole phase,
the neck, driven by the top plate shows a kind of linear
amplitude increase like the inclosed air, because it is directly connected to the top plate and so constantly driven
by the top plate like the inclosed air. The neck has the
highest resonance spectrum as expected.
So from the mentioned guitar parts like the top plate, the
back plate, the ribs, the inclosed air and the neck, only
the top plate comes in instantaneously with a hard attack
and so is the only part of the guitar responsible and able
to have an attack. So when hardening the attack of guitars, the most efficient way is to make the top plate vibrate as freely as possible as is known by guitar builders
very well. Still this top plate does not make the guitar
sound, because here we would have just a string acting
on a wooden plate. It sounds like a plucked stringed
instrument, but what makes the classical guitar suitable
for playing music, where a deep and prominent sound is
needed, like the 19th century concert hall music or like
contemporary classical music, the other parts of the guitar
are needed. The ribs and the back plate come in about at
the same delayed time point compared to the top plate and
support their middle frequency region. Here, the color
of the guitar is produced. The manipulation of these parts
lead to a manipulation of the fine structures of the radiated guitar tone. The inclosed air adds a kind of pumping
to the sound, first of all, because it is low in frequency,
and then because is shows a linear increase in amplitude.
Finally, the neck adds a high frequency sound needed for
some of the brilliance of the instrument.
Rolf Bader
F = T sin
(3)
dy
,
dx
(4)
where y is the displacement of the string in the normal direction of the top plate and x is the direction of the string.
So the force of the string acting periodically on the top
plate is described as a first-order differential equation.
When applying this in a model of the whole geometry of
the guitar [1], the resulting response of the guitar top plate
misses all overtones and is just a low frequency sound,
still in the periodicity of the string. When taking the system under closer geometrical examination is shows up,
that the force acting on the top plate need at least one
second term, the force from the strings differential equation
d2 y
d2 y
= c2 2
2
dt
dx
(5)
(6)
the accelerating force FA of Newtons law and the backdriving force FB of the string caused by its curvature at
each string point (dividing both sides by the mass m).
So this differential equation of the string claims, that at
each point of the string, the force FA of acceleration of
this point in the normal vector direction of the top plate
is caused by the second derivation of the string displacement at that point, in other words according to the curvature at that point, as a linear slope of the string has zero
second derivation and so no backdriving force.
Figure 2: Development of the initial transient of the different guitar parts. From top to bottom: top plate, back
plate, ribs, neck, inclosed air. The top plate comes in
first with a hard attack followed by the inclosed air and
the neck both showing more or less linear amplitude increase. The back plate and the ribs come in later with a
sudden boost.
Now this holds for all points of the string and so also
for the bridge point of the string, where this backdriving force now acts on the bridge. When modelling both
forces, the first and second derivative of the string at the
bridge point, the missing overtone structure is there. The
first order differential equation transmits the low frequencies and the second order differential equation the high
frequencies from the string to the bridge. This nonlinear
term of second order is not needed when driving a guitar
bridge artificially by a force acting directly in the normal
vector direction of the bridge, because this could be modelled as a point source, as zero-dimensional. The string
can be looked at - still simplified - as a one-dimensional
geometry and so needs a second nonlinear coupling term.
Rolf Bader
References
[1] Bader, R.: Computational Mechanics of the Classical
Guitar, Springer 2005.
[2] Bader, R.: Physical model of a complete classical
guitar body, Proceedings of the Stockholm Music
Acoustics Conference 2003, R. Bresin (ed.) Vol. 1,
121-124, 2003.
[3] Woodhouse, J.:The Transient Behaviour of Guitar
Strings, Proceedings of Stockholm Musical Acoustics Conference 2003, 137-140, 2003.
[4] Woodhouse, J.: Plucked Guitar Transients: Comparison of Measurements and Synthesis, Acustica 90/5,
945-965, 2004.
5 Conclusion
Most of the fine structures governing the sound of the
classical guitar is produced by nonlinear effects like the
coupling of bending and in-plane motion of the plates or
like the nonlinearity inherent in the geometry of the guitar itself. Additionally, the coupling of the string to the
bridge and the guitar top plate implies a nonlinear term
of the second-order differentiation of the string displacement with respect to space at the bridge point.