Session 01 Violins
INVITED
The bowed string as we understand it today
Woodhouse, Jim
Cambridge University, Engineering, Cambridge, United Kingdom
The vibration of a bowed string has been studied since the 19th century, and today this problem is the only example of vibration excited by friction which can claim to be well understood. Today's theoretical models embody many of the complicating features of real strings, instruments and bows, and detailed comparisons with experimental studies are allowing fine tuning of the models to take place. The models can be used to explore questions directly relevant to instrument makers and players This lecture will review the history of bowed-string research, and highlight recent developments and current issues. Probably the main unresolved question at present concerns an accurate description of the frictional behaviour of rosin, since recent results have shown that none of the friction models used in the existing literature is entirely credible.
INVITED
Generalized normal mode violin acoustics
G Bissinger
East Carolina U, Physics, Greenville, NC, United States
A constantly growing violin normal mode database presents
significant opportunities to relate specific modal properties such as the 1st
corpus bending mode frequencies to violin quality. Perhaps more interesting however
is the prospect of extracting general aspects of violin sound related to quality
classes such as "good" or "bad". Frequency-related trends
for various measured properties, viz., total damping, radiation efficiency/damping,
and internal damping, are of fundamental interest because of their connection
to the vibratory energy expenditure of the violin. For example, (aggregate) violin
internal damping obtained by subtracting radiation damping from total damping
was surprisingly insensitive to violin quality, while the parent dampings were
not. Because damping trendlines for each energy loss path are independent of excitation
particulars, they provide insights into the spectral balance of radiation from
the violin while sidestepping the role of the intermediate bridge-corpus energy
transfer, i.e., a "bridge-less" violin. Computing the fraction-of-vibrational-energy-radiated
from the radiation-total damping ratio up to 8 kHz reveals a mechanism related
to the critical frequency for a "good" violin radiating more in the
3 kHz region relative to a "bad" violin. Finally some aspects of f-hole
vs. structure contributions to the far-field radiation will be presented. ORAL The form of the violin has prevailed as the standard
for the violin, viola and violoncello, but not for the double bass. Surveys
reveal that while some tendencies exist among instrument makers and musicians,
opinions about the sound of the two main types of basses, flat- and round-backed,
often conflict. Several new double basses of nearly identical form and extremely
similar materials were created especially for this study, presenting a unique
opportunity to define acoustical characteristics of flat-backed and round-backed
basses. The pairs of instruments differ, in theory, only in the form of the
back. Laser vibrometry, analysis of audio recordings made in an anechoic chamber,
and listening test surveys were used to delineate differences between the types
concerning radiation characteristics and timbre. A special flat back for an
existing violoncello was also made, tested and compared to a similar instrument
of the standard form. Room acoustics and psychoacoustical considerations are
discussed along with experimental results to offer an explanation about why
the double bass is the only modern bowed instrument in use that has a flat back. ORAL TV-holography has been used to study operating deflecting
modes in seven violins including an old Italian. Two of these instruments were
made of plywood with a special bar design, the patented "Hagetrø
violin". Mode shapes, vibration phase, -frequencies, levels and Q-values
were recorded for frequencies under 1400Hz. Generally corresponding modes of
the seven violins looked similar, though differences were detected, especially
in phase mappings of the vibrations. Combinations between modes were observed
as the rule. The C3 and C4 modes appear as breathers in some instruments and
as non-breathers in others. The Hagetrø violins were both breathing in
these modes. All instruments had double or triple resonances where distinct
modes appeared with similar vibration patterns at two or three maxima. This
coupling effect was somewhat clearer in the higher quality instruments. Vibration
velocity plots show clear differences in vibration levels among the violins.
Playing tests in an echo free room indicated that the more intense sounding
instruments are those with higher velocity levels. The modes were more damped
in five varnished than in two unvarnished instruments. POSTER The wood material for violin Bows from the Pernambuk
tree (Guilandina echinata) has large individual variations in the density. For
high quality bows, the higher densities are preferred, but since the Pernambuk
is rather rare and expensive, it is of interest to investigate if it is possible
to compensate for density variations in the wood material, at the same time
as the first eigenfrequencies, the total mass and location of the center of
gravity of the bow is unchanged. The mechanical calculations of the bow is made
with a numerical model in the finite element program ANSYS, and in this first
approach, the calculations considers only the elastic properties of the bow,
damping is not included. In order to perform automatic optimization of the geometry
of the bow, the FEM-program ANSYS is coupled to an external optimization routine.
The stochastic Simulated Annealing method (SA) is used as optimization routine,
and the routine has showed good performance in similar problems with several
local minima. POSTER In order to validate and develop theoretical models
of the action of a bowed string, careful experimental work is needed. It is
already known that current models predict behaviour which is qualitatively plausible,
but can they predict accurately the details of the transient response to specific
bowing gestures? A computer-controlled bowing machine has been developed which
allows reproducible gestures to be made, in which the velocity and force waveforms
can be specified. This allows systematic experiments to be made in which a portion
of parameter space is explored by varying parameters in small steps. These results
can be compared directly to corresponding simulated results, and the identical
automatic processing of the string transients can be applied to both so that
the comparison is fair. This comparison reveals that some aspects of behaviour
are quite well captured by current models, but that other features are not reproduced.
Careful study of the results can guide further development of the theoretical
models. ORAL
"Is it possible to make sound copies of violins?"
To answer this question one needs to know the ingredients which make two violins
being judged by musicians to be aurally identical. Using a questionnaire among
50 musicians verbal attributes related to the sound characteristics of violins
were collected and ranked. Subjective listening tests were carried out to check
which verbal attributes are suitable and do evoke the same meaning in different
persons. Useful attributes for the ranking could be found using statistical measures
for consistency and concordance. Additionally, it was found that the measurement
uncertainty of frequency response measurements today is smaller than the analysis
resolution of the human ear in regard to frequency and amplitude in the frequency
range important for the hearing sensation and for room acoustic situations typical
for a violin maker's workshop. The results give strong indications that the psychoacoustic
characteristics of the human ear and the vibroacoustic characteristics of the
violin body in combination with the attributes judged to be important for violin
sound quality will in principle allow for the idea of making sound copies of a
given violin. ORAL We investigate the influence of vibrato and noise on
the perception of the tonal quality of a violin in two ways. First, we consider
recorded sound of both high and poor quality violins. By continuous repetition
of single period waveforms, we remove the frequency, amplitude, and noise fluctuations
introduced by the performer playing the instrument, whilst retaining the sampled
spectral information relating to resonances of the particular violin being played.
Audio demonstrations demonstrate the importance of the fluctuations in frequency,
amplitude and spectral timbre in identifying the sound as that of a violin.
We therefore argue that such fluctuations must be equally important in any subjective
assessment of the quality of a violinīs tone. To investigate such correlations,
we have developed computational models, which enable the listener to judge the
influence of vibrato amplitude and frequency on the tone of an idealised multi-resonant
violin. In particular, we also demonstrate how the influence of vibrato depends
critically on the Q-values and density of the vibrational modes excited. Such
factors are clearly correlated with the "quality" of the wood used
in a violinīs construction and the thicknesssing of plates by the luthier. ORAL Raman and Schelleng analysed waveform properties of
the bowed string as function of bow speed, bow force, and bow position (the
distance between bow and bridge). Schelleng also described spectral changes
caused by alterations of the bow force alone. This phenomenon was later explained
by Cremer, and referred to as "rounding of the Helmholtz corner".
In this study it is shown that of the remaining two parameters, it is the bow's
speed rather than its position that bears a potential of changing the shape
of the string's spectral envelope. This contrasts to the popular belief that
by bringing the bow closer to the bridge, the sound automatically becomes more
brilliant. POSTER In the frequency range of maximum sensitivity of the
ear the so-called BH-hill can be found. The BH-hill is not confined to the violin
bridge only. A normal bridge and a bridge plate (a bridge with only feet tips
but no cutting outs as heart, ear etc) give mainly the same BH-hills. The same
bridge on a Stradivarius violin and a new violin gives differently shaped BH-hills. ORAL We apply a mixture of unsupervised learning and statistical
methods to find and illustrate similarity and dissimilarity factors in the timbre
of violin voices. Our data come from the AMATI sound database that contains
digitized recordings of 70 musical instruments presented at the Henryk Wieniawski
10th International Violinmakers Competition in Poznan, Poland, 2001. The raw
sound data served to extract various sets of features, including harmonic based
parameters (e.g. brightness, tristimulus, even and odd harmonics content), spectral
parameters (e.g. energy, moments of various order), mel and linear scale cepstral
coefficients, spectral envelope features (maxima and minima) and human ear auditory
model features (the Lyon's model). We demonstrate machine ability to perceive
the differences of violin timbre by applying multidimensional scaling (MDS)
paradigm to the results of both classic and enhanced clustering methods. The
MDS-based visualization leads to better perception and comprehension of the
analyzed multi-dimensional space of the timbre characteristics. The ultimate
MDS result is finally enhanced by the introduction of Voronoi diagrams, which
can be displayed in form of both planar and spherical maps. POSTER
We present a database AMATI, entirely devoted to the violin
voices. The instruments recorded are master quality and represent international
schools of violinmaking. They were competitors of the 10th International Henryk
Wieniawski Violinmaker Competition, held in Poznan in autumn 2001. There are 17
000 sound files gathered in the database. The collection of sounds comprise material
similar to the one that the jury of musicians examined during the audition. It
includes open string bowed and pizzicato sounds, the entire range of notes across
a chromatic scale on each string, a range of notes of diatonic scale and a fragment
of J.S. Bach's work. For each instrument the basic information is provided, including
a photo, an emblem, the violinmaker's nationality, etc. The averaged scores of
musical jury judgements constitute the unique and important element of the database,
making search for relationships between objective and subjective measure possible.
The database contains harmonics based parameters for each sound. Parameter values
may be visualized and compared in multiple configurations using a variety of charts
created on-line. Signal waveforms, spectra and spectrograms are also available.
Numerical values of parameters may be exported and used for further study. The
database is an open source product (My SQL) and may be distributed among researchers. POSTER The quality evaluation of elements assessed for bowed
instruments making is important for forest economists, wood businessman as well
as for music instrument makers. In order to get the more complex data for violin
wedges evaluation, to the trinity of characteristics of physical acoustics,
measured in previous experiments (the density, modulus of elasticity, acoustical
constant) the measurement of logarithmic damping decrement was added. After
construction and evaluation of experimental violins made from the samples investigated,
the more close linkage between materials characteristics and quality of instruments
should be expected. POSTER Violinists often claim that tilting the bow provides
greater brilliance. By tilting, the effective width of the bow hair is reduced
and the bow force distribution across the bow hair ribbon is changed. Considering
that the width of the bow hair of a violin bow is roughly 1/32 of the string
length (about 10 mm), and that the relative bow-bridge distance in playing typically
varies between 1/8 and 1/32, an effect of the width of the hair on the slip-stick
process seems reasonable. Pitteroff has reported simulations and measurements
showing that the slipping intervals become progressively shorter as the width
of the bow hair ribbon is decreased. [Pitteroff & Woodhouse et al, acta
acustica - Acustica 84, 929-946]. However, the effect, which mainly was attributed
to a faster transition from stick to slip at release, was small. In this study,
evidence gained in experiments using a bowing machine are presented, showing
that a decrease of the width of the bow hair may boost the string spectrum considerably
for higher harmonics. A gain in partial amplitudes of 3 - 6 dB has been observed
above harmonic #20. Besides increased brilliance, it is clear that there are
several other reasons for violinists to tilt the bow. For example, tilting the
bow hair facilitates gentle note attacks due to a gradual buildup in string
contact. POSTER An important element of a violin is its body, which
filters vibrations that propagate from the string through the bridge. In real-time
synthesis of a violin, there is some difficulty in modeling the body because
of a tradeoff between accuracy and computational cost. If all the resonances
of the body are accounted for by modeling each one with its own pair of filter
poles, the computational cost is too high. On the other hand, one cannot implement
too few filter poles and neglect the large number of resonances, because the
complex filtering of the body contributes strongly to the characteristic timbre
of the violin. The complexity of the spectrum of a violin's body is due to the
complexity of the shape of the instrument, which is the result of centuries
of developments. In the 19th century, Savart proposed a simplification of this
shape building a trapezoidal violin. Documents of the period report that, even
if the violin was never adopted, its sonorities were perceptually equivalent
to the ones of the traditionally shaped violin. In this paper we propose a digital
implementation of Savart's trapezoidal violin using a three dimensional waveguide
mesh. The dimensions of the mesh are chosen in order to match the dimensions
of the instrument as built by Savart. Digital waveguide filters have been used
to simulate damped boundary reflection of the acoustic waves against the violin
body. Advantages and limitations of Savart's trapezoidal violin are discussed. ORAL The interaction of a bow exciting a string is a complex
phenomenon that belongs to the larger field of frictional induced vibrations.
In the past, theoretical models of the bow-string interaction assumed that the
frictional force at the contact point depends only on the relative sliding velocity
between the bow and the string. Recently, Smith and Woodhouse observed that
rosin exhibits plastic deformation at the contact point. For this reason, they
proposed a model called plastic, in which friction has a dependence on the temperature
variations in the interfacial rosin layer. This model exhibits an hysteresis
loop in the velocity versus force plane. In robotics and haptics, in order to
better predict friction phenomena, dynamic friction models have been proposed,
in which the dependence of friction on the relative sliding velocity is given
by a differential equation. In this paper we apply the state of the art dynamic
friction models. i.e. the elasto-plastic friction models, to the simulation
of a bow exciting a string, proposing an accurate yet efficient numerical implementation.
Comparisons between Smith and Woodhouse's plastic model and the dynamic friction
models are described, which show that an accurate choice of the control's parameters
in the dynamic models allow to obtain an hysteresis loop in the friction versus
velocity plane similar to the one produced by the plastic model. ORAL Bowed string physical models have achieved a level of
completeness that allows their performance to be favorably compared to that
of their real instrument counterparts. These comparisons are facilitated by
the use of refined bow controllers that detect all the subtle changes in motion
and force that are experienced by a bow while in contact with a string and give
expressive playing its characteristic sound. In this paper we use a real-time
bowed string physical model and a wireless bow controller to reproduce the bow
strokes that are most fundamental to the right hand technique of an accomplished
bowed string player, such as legato, detaché staccato, spiccato, and
balzato. We discuss the integration of the two components of these experiments
and illustrate how the bow controller is used to control the physical model
of the violin in order to faithfully reproduce these strokes. Moreover, we compare
the range of input parameters that determine these strokes in the model with
the values for these parameters measured on real violins, showing how synthetic
instruments may present the same playability regions as real instruments. POSTER We present results of analyses carried out for violins
that were submitted to the 10th International Henryk Wieniawski Violinmakers
Competition. In the analyses we were especially interested in reconstructing
the relationship between some pre-defined characteristics of the instruments
and the verdict reached by the jury. Taking into account the following criteria:
volume of sound, timbre of sound, ease of sound emission, equal sound volume
of strings, accuracy of assembly and individual qualities of the instrument,
the jury ranked the instruments from the best down to the worst. On the base
of this ranking we attempt to infer a preference model that is supposed to re-construct
the preference of the jury. We use inductive supervised learning methods that
include a preference-modeling tool called Dominance-based Rough Set Approach.
The analysis starts with constructing a rough approximation of the preference
relation underlying the final ranking. This allows inducing decision rules in
terms of criteria considered by the jury but also in terms of other criteria,
including various acoustic characteristics of the violins. In both cases, the
rules can serve to: (i) explain the decision policy of the jury, (ii) build
new rankings for new sets of violins, either evaluated on the jury's criteria
or characterized by acoustic features. The approach is especially useful in
discovering subsets of acoustic features that are relevant to producing rankings
of violins. ORAL In the strive of understanding how different parameters
affects the vibration properties and the characteristics of the sound emanating
from a violin, i.e. what makes a good violin good, numerical methods as FEM
(finite element method) and BEM (boundary element method) are used. Numerical
models of whole violins and/or part of it is created and studied. Crucial for
the results from these studies is the correctness of the input data for the
numerical analysis. One important group of input data is the wooden material
parameters for the part of the violin subjected to analysis. In this study a
new method for determining these important material parameters for blanks for
violin tops is proposed. In the proposed method a FEM-code is linked together
with a stochastic optimization algorithm in order to, in an automatic fashion,
determine the material parameters. The method requires the geometrical dimensions,
the density, and measured normal modes for the blank and it consider the fact
that the Young's modulus in the longitudinal direction varies in the radial
direction. POSTER We present a new system to measure the position and
force parameters most relevant to violin bowing technique. It is stable and
reliable, and as it allows players to perform without impediment and consists
of minimal equipment, it is also suitable for use in professional performance
settings. This measurement system, implemented on a carbon fiber bow, relies
on an electromagnetic field sensing technique for detecting both transverse
bow position and bow-bridge distance and foil strain gauges for downward force
measurement. The position data is collected using an antenna that is located
behind the bridge and mounted from the tailpiece of an acoustic violin. The
strain gauges are permanently mounted around the midpoint of the bow stick,
and the force data is collected and sent to a remote computer via a wireless
transmitter mounted at the frog. The resultant bow remains wireless, and the
placement of the sensors and electronics ensures a balance point that remains
within the normal range for traditional bows. We describe several experiments
illustrating the usefulness of this system for continued studies of bowing parameters,
as well as inspection of musical performance style and pedagogical applications. ORAL The 3x3 admittance matrix at the centre of a cello bridge
has been measured directly, using a miniature impulse hammer and laser vibrometer.
This contrasts with earlier work, in which the matrix was determined (for a
violin) by ingenious but indirect means. Data exhibiting good coherence and
correct reciprocal behaviour has been obtained up to 3 kHz (corresponding to
9 kHz on the violin). Results will be shown, and the interpretation will be
discussed of some unusual features appearing in the admittances for excitation
parallel to the strings. Out-of-plane resonances of the bridge are seen, together
with an apparently new vibration mode of the cello structure at about 400 Hz.
An acoustical comparison of flat- and round-backed double
basses
A W Brown
University of Music and Dramatic Arts Vienna, Institut
fuer Wiener Klangstil, Vienna, Austria
Operating deflecting modes of five conventional and
two unconventional violins recorded with TV-holography
A Buen¹, O J Løkberg²
¹Brekke & Strand akustikk as, Building acoustics,
Oslo, Norway; ²The Norwegian University of Science and Technology, Institute
of Physics, Trondheim, Norway
Numerical optimization of violin bows with varying densities
of the wood material
P Carlsson, M Tinnsten
Mid Sweden University, Information Technology and Media,
Ostersund, Sverige
Experiments with an automatic bowing machine
P M Galluzzo, J Woodhouse
Cambridge University, Engineering, Cambridge, United Kingdom
Psychoacoustic investigations on the possibility of
aurally identical violins
P Geissler¹, O Martner¹, C Zerbs¹, M Schleske²
¹Müller-BBM, Schalltechnisches Beratungsbüro,
Planegg, Germany; ²Meisteratelier für Geigenbau, Munich, Germany
The influence of vibrato and noise on the assessment
of violin tone
C Gough
University of Birmingham, School of Physics and Astronomy,
Birmingham, United Kingdom
Bow speed or bowing position - which one influences
spectrum the most?
K Guettler¹, E Schoonderwaldt², A Askenfelt²
¹Norwegian Academy of Music, Oslo, Norway; ²KTH,
Dept. of Speech, Music and Hearing, Stockholm, Sweden
The BH-hill and tonal quality of the violin
E V Jansson
KTH, TMH, Stockholm, Sweden
A couple of forces (two forces in opposite directions) one at each bridge foot
gives a BH-hill, both for a normal bridge and a bridge plate. A momentum applied
at the central line of the top plate gives no BH-hill.
The BH-hill is much affected by the distance between the two feet, i.e. between
the two forces of the couple. The cross-stiffness of the top plate also gives
a large influence.
The importance of the bridge hill will be demonstrated by playing two violins,
one with a normal BH-hill and another violin with no BH-hill.
Thus it will be demonstrated that the BH-hill is very important and that it
is mainly set by the bridge feet distance and the top plate properties at the
bridge feet.
Unsupervised machine learning methods in timbral violin
characteristics visualization
E Lukasik, R Susmaga
Poznan University of Technology, Institute of Computing
Science, Poznan, Poland
AMATI: A multimedia database of violin sounds
E Lukasik
Poznan University of Technology, Institute of Computing
Science, Poznan, Poland
Logarithmic damping decrement as the characteristic
of physical acoustics for quality estimation of violin wedges
E Rajcan, S Urgela, M Culik
Technical University Zvolen, Zvolen, Slovakia
Effect of the width of the bow hair on the violin string
spectrum
E Schoonderwaldt¹, K Guettler², A Askenfelt¹
¹KTH, Dept. of Speech, Music and Hearing, Stockholm,
Sweden; ²Norwegian Academy of Music, Oslo, Norway
Modeling Savart 's trapezoidal violin using a digital
waveguide mesh
S Serafin¹, F Fontana²
¹CCRMA, Music, Stanford, United States; ²University
of Padova
Department of Information Engineering, Padova, Italy
Bowed string simulation using an elasto-plastic friction
model
S Serafin¹, F Avanzini², D Rocchesso³
¹CCRMA, Department of Music, Stanford, United States;
²University of Padua, D.E.I., Padua, Italy; ³University of Verona,
Computer Science, Verona, Italy
Bowed string physical model validation through use of
a bow controller and examination of bow strokes
S Serafin¹, D Young 2
¹Stanford University, CCRMA, Stanford, CA, United
States; ²Massachusetts Institute of Technology, Media Lab, Cambridge, MA,
United States
Inferring decision rules from jurys' ranking of competing
violins
R Slowinski, J Jelonek
Poznan University of Technology, Institute of Computing
Science, Poznan, Poland
Determination of important wood properties for blanks
for violin tops by the use of numerical optimization
M Tinnsten, P Carlsson
Mid Sweden University, Information Technology and Media,
Ostersund, Sverige
Wireless sensor system for measurement of violin bowing
parameters
D Young
Massachusetts Institute of Technology, Media Lab, Cambridge,
MA, United States
The admittance matrix of a cello
J Woodhouse, P Courtney
Cambridge University, Engineering, Cambridge, United Kingdom
