Radiation characteristics of multiple and single sound hole

vihuelas and a classical guitar

Rolf Badera)
Institute of Musicology, University of Hamburg, Neue Rabenstrasse 13, 20354 Hamburg, Germany

(Received 29 October 2010; revised 24 February 2011; accepted 27 February 2011)

Two recently built vihuelas, quasi-replicas of the Spanish Renaissance guitar, one with a small
body and one sound hole and one with a large body with five sound holes, together with a classical
guitar are investigated. Frequency dependent radiation strengths are measured using a 128 micro-
phone array, back-propagating the frequency dependent sound field upon the body surface. All
three instruments have a strong sound hole radiation within the low frequency range. Here the five
tone holes vihuela has a much wider frequency region of strong sound hole radiation up to about
500 Hz, whereas the single hole instruments only have strong sound hole radiations up to about 300
Hz due to the enlarged radiation area of the sound holes. The strong broadband radiation of the five
sound hole vihuela up to about 500 Hz is also caused by the sound hole phases, showing very con-
sistent in-phase relations up to this frequency range. Also the radiation strength of the sound holes
placed nearer to the center of the sound box are much stronger than those near the ribs, pointing to
a strong position dependency of sound hole to radiation strength. The Helmholtz resonance fre-
quency of the five sound hole vihuela is influenced by this difference in radiation strength but not
by the rosettas, which only have a slight effect on the Helmholtz frequency.
C 2012 Acoustical Society of America. [DOI: 10.1121/1.3651096]
V

PACS number(s): 43.75.Gh, 43.35.Sx, 43.40.Rj [NHF] Pages: 819–828

I. INTRODUCTION instrument seems to never have really been played. Because

of its string length, the standard tuning could not be applied,
In the present paper, three instruments are compared
and so the known repertoire of the vihuela cannot be per-
in terms of the role of the sound holes to the overall radia-
formed on it. Most likely it was an example instrument for
tion behavior of the instrument, a modern classical guitar
builders to examine. The five sound holes have different sizes,
(Lauenhardt and Kobs), a small sized vihuela with one sound
where the center and the two lower holes (near the bridge) are
hole (Yaish), and a guitar-like sized vihuela with five sound
larger than the upper holes. Another vihuela was built 1620
holes (Herb).
and found in Quito, the capital of Ecuador. Its string length is
The vihuela, a Spanish Renaissance guitar-like instru-
72.7 cm coming closer to that of the modern guitar.
ment, has only survived with three known instruments, and
During the last about 30 years in the field of historical
only seven historical song books for vihuela are known. Still,
performance practice, musicians and instrument builders
many pictures show the instrument where we find many
have tried to re-vitalize and perform Renaissance and
sizes, from a soprano to a contra-bass instrument in some
baroque music as close to historical sources as possible. Still,
variations of its geometry and ornamentation. Additionally,
some practical changes have been made, especially in instru-
the Sevillan guild order from 1502 describes the instrument
ment building, to fit the needs of modern performers. So
roughly.1,2 In all cases, it is made of ribs and so clearly dis-
within the last decades, especially lute makers but also guitar
tinguished from a lute that has a curved back. Still its most
builders were asked by performers to rebuild vihuelas. As
popular tuning is like that of the lute, G-c-f-a-d0 -g0 , which is
then only two remaining instruments were known (the third
very similar to a modern guitar tuning, only a minor third
was only discovered recently), the instruments built were
higher and with a major third interval between the third and
mostly a mixture of the original instruments and modern
fourth string instead of the fourth between these strings used
needs. The Herb vihuela is such an example, where the five
with modern guitars. From the three remaining instruments,
sound hole set was used in a sound box with the size of a
the most famous is the so-called Guadelupe instrument,
guitar. As none of the remaining instruments can be played
named after an engraving pointing to the cloister Guadelupe
anymore, builders often tried to shape the instrument sound
in the Spanish Extremadura region, and is now located in the
as a mixture of a guitar and a lute sound. If this is historically
Musee-Jacquemart-Andre in Paris. This instrument has five
correct or not cannot be verified any more.
sound holes arranged like the five points on a dice. Its size is
The role of the five sound holes compared to instru-
larger than that of a modern guitar with a string length of
ments with only one sound hole has often been debated but
79.8 cm (65 cm is a modern guitar string length).3 This
not studied systematically with the vihuela yet. Still, the
behavior of sound holes with wooden boxed instruments has
a)
Author to whom correspondence should be addressed. Electronic address: been studied extensively, often using a two-mass model like
[email protected] Le Pichon et al. (1998),4 who use the Helmholtz and the

J. Acoust. Soc. Am. 131 (1), Pt. 2, January 2012 0001-4966/2012/131(1)/819/10/$30.00 C 2012 Acoustical Society of America
V 819

Downloaded 12 May 2013 to 131.104.62.10. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
two-mass model to predict the plate and volume velocities. guitar from Hans H. Herb a luthier from Erlangen, Germany
Multi sound hole cases have been studied by Peekna and built in 1988. All three instruments are shown in Fig. 1.
Rossing (2005).5 They investigated psalteries with different The Yaish vihuelas’ top plate is made of spruce, the ribs
amount of sound holes, calculating the Helmholtz resonance and back are Rio rosewood with ribs of European holly (ilex
f as a sum of the sound hole radius aj and area Aj, the volume aquifolium) and ebony, and the neck and peg box are made
of the air inside the box V, the soundboard thickness at the of sycamore. It is six stringed double-choired with two
sound hole location Lj and the speed of sound c like strings tuned to the same pitch (so 12 strings all together)
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi with free vibrating string length (diapason) of 57 cm. The
u X n frets are made of movable strings as historically used (here
c u
t1 Aj
fA ¼ : (1) nylon instead of gut is used with both strings and frets). The
2p V j¼1 16aj =3p þ Lj
body has a length of 38.7 cm a larger “belly” of 24.9 cm, a
smaller “belly”’ of 21.5 cm, and a waist of 19.4 cm (see Fig.
The interaction between this air resonance and the top plate 1 for the small waist construction). The neck is 27.6 cm long
resonance fp result in two frequencies f1 and f2 as combina- and has a larger width at the sound box of 6.2 cm and a
tional vibrations according to the formula f12 þ f22 ¼ fA2 þ fp2 . smaller width of 4.8 cm at the peg box end. The following
They find excellent results for single hole cases which get peg box then is 13.5 cm long and continues the necks’ width.
worse for multiple sound holes. The authors claim the reason The vihuela has a ribs height of 8 cm. The sound hole has a
to be the neglected damping of air at the sound holes which diameter of 7.1 cm. The rose inserted into the sound hole
may differ in terms of sound hole size. In another experiment reduces the area to about 65% of the total area. This was
of Peekna and Rossing,6 the size of the sound hole was var- measured by taking a picture of the rose and using image
ied, and the resulting brightness of the sound radiated from analysis techniques to determine the area of the rose in the
the instrument was judged in terms of a desired brightness sound hole. Using the same technique, the total area of the
for a psaltery by listening tests. With sound hole area too top plate was found to be 758 cm2 and therefore its air vol-
small, the instrument sounded too dull and less bright, and a ume is 6065 cm3.
best sound hole area was found to meet listeners preference. The Herb vihuela has a spruce top plate, and the ribs
As a change of sound hole area is reported to change the fre- and back are made of cherry wood. As there is no known fan
quencies of the lower two body resonances, the change of bracing from Renaissance times, the instrument has bars at
brightness is associated with these frequency shifts. the top and back plate. The top plate is between 1.9 and 2.1
For single tone holes a standard formula to calculate its cm in thickness, the back plate is about 2.8 cm. Is is also
frequencies is given by Fletcher and Rossing7 like double-choired with a vibrating string length of 65 cm like
rffiffiffiffiffiffiffiffiffiffiffi the modern guitar. This measure was intended by the builder
c 0:85r as asked by the ordering musician. The body has a length of
fA ¼ ; (2) 48 cm, a larger “belly” of 30.6 cm, a smaller “belly” of 24.0
2p V
cm, and a waist of 21.9 cm, again not showing a large waist
where r is the sound hole radius and V is the air volume in (see Fig. 1 for a comparison of the two vihuelas with a mod-
the box. ern guitar). The neck has a length of 29.4 cm with a larger
McPherson8 has a patent proposing to shift the sound width of 6.8 cm at the sound box and a smaller width of 5.5
holes of guitars closer to the body edge to increase sound cm at the peg box end. The peg box then is 20.4 cm long.
quality, still no evidence is presented for this change. The rib height of the vihuela is 7.3 cm. This vihuela, like the
Strong et al.9 investigated guitars using acoustic holog- historical Guadelupe instrument has five sound hole ordered
raphy. They found a strong radiation from the sound hole like the dots on a dice of the number five, where the two
dominating the overall guitar sound radiation in the very holes closer to the bridge have the same size as the hole in
bass region.
The sound holes of violins have been studied, too.
Weinreich10 discusses the dipole and monopole radiation
characteristics of the instrument using measurements. He
finds strongest monopole radiation at the Helmholtz reso-
nance as expected and a lowering of the air radiation below
this resonance due to phase differences between the two
f-holes. Bissinger et al.11 measure the radiation of the second
violin air mode A1, which has a nodal line right in the
middle through the f-holes resulting in a very low radiation
of this mode due to phase cancellations.

II. GEOMETRIES OF THE INSTRUMENTS

Two vihuelas were used in the investigation, a small (alt)
sized one built by the luthier Avital Yaish in 1997 in FIG. 1. The three instruments used; from left to right: Lauenhardt and Kobs
Jerusalem, Israel, and a larger one about the size of a classical (LK) classical guitar, Herb vihuela, and Yaish vihuela.

820 J. Acoust. Soc. Am., Vol. 131, No. 1, Pt. 2, January 2012 Rolf Bader: Multiple vihuela sound hole radiation

Downloaded 12 May 2013 to 131.104.62.10. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
the middle with diameters of 7.2 cm each. The two holes
closer to the neck are smaller, both with a diameter of 5.4
cm. Although the roses are as elaborated as with the Yaish
instrument, the arabesque is thinner and therefore the effec-
tive radiating area of the sound holes is again smaller. As
with the Yaish rosettas, using image processing the radiation
area was again found to be 65% of the total area. Using the
same technique, the total area of the top plate was found to
be 1|137 cm2 and therefore its air volume is 8304 cm3.
The Lauenhardt and Kobs classical guitar has a spruce
top plate and palisander back plate and ribs. Its body is 49
cm long, the larger “belly” is 36.9 cm and the smaller one
29.2 cm wide. The waist is 24.4 cm with a rib height of 9.5
cm. The sound hole has a diameter of 9 cm. The string length
is traditionally 65 cm. The guitar has normal Torres fan brac-
ing. It has a top plate area of 1|304 cm2 and its volume is
12|385 cm3.

III. METHOD
All three instruments were placed 3 cm in front of two
microphone arrays of 121 microphones. The array recorded
two seconds of plucked tones with a sampling frequency of
48 kHz per microphone simultaneously. The microphone
array recordings were Fourier transformed, and the sound
field for each frequency at the microphone positions was
back-propagated to the instrument surface using the minimum
energy method as proposed in Bader12 (see this reference also
for technical details about recording equipment), a kind of an
equivalent source method (for further literature about equiva- FIG. 2. Experimental setup of the microphone array measuring the guitar.
lent source methods, see Ref. 13, near-field acoustic hologra-
phy,14 Helmholtz least-square method,15 and a review paper
on the field in Ref. 16). Figure 2 shows the measurement the directivity value. If a monopole is assumed, to which the
setup. Here opposite each microphone, a radiation point is sound hole radiation of the guitar does come close, a  0.
assumed. Each radiation point is a complex monopole source With a more complex radiation pattern, we would expect the
with a directivity value a, where a ¼ 0 is a perfect monopole, directivity to be more narrow and a increases.
and a > 0 narrows the radiation with respect to the top plate A linear equation solver is then back-propagating the
normal vector. The recorded pressure pjm at the jth micro- measured complex amplitudes to the radiation points on the
phone is simply assumed to be the summed pressure values of guitar top plate using the radiation matrix R. As R depends
all radiating points pig on the geometry (vihuela or guitar top on a, which is arbitrary, the solution process is repeated for
plate) taking phase changes and radiation loss in a radiation different values of a. For each solution, the reconstruction
matrix Ri,j into consideration like energy is calculated by integrating the powered pressure val-
ues. It appears that for the correct reconstruction, this recon-
X
N
struction energy has a minimum. This behavior can also
pjm ¼ pig Rij : (3) intuitively be understood easily. If a is very high, only radia-
i¼1
tion into the microphone opposite to the radiation point is
This matrix Rij takes phase changes into consideration assumed and the reconstruction would simply be the mea-
through the wave vector k and the distance rij between surement. But then no superposition of radiation points at
microphone i and radiation point j like the microphones would take place, and each radiating point
would need to supply all the energy needed for the opposite
1 ıkrij microphone. By reducing a, the radiation points contribute
Rij ¼ e : (4) also to the neighboring points and therefore reduce the
Cij
reconstruction energy. Visually, this can be seen as a contin-
The amplitude drop uous sharpening of the reconstruction when decreasing a
compared to the blurred recorded sound field. On the other
Cij ðaÞ ¼ r ij ð1 þ að1  bij ÞÞ (5) side, if a is physically small, too much influence of neighbor-
ing points is assumed, and the linear equation solver tries to
also depends upon the radiation angle b, defined as b ¼ 1 in balance the recordings, which it only can do by assuming
the normal direction and zero in the direction 690. a  0 is unreasonable large amplitudes. So the solution blows up

J. Acoust. Soc. Am., Vol. 131, No. 1, Pt. 2, January 2012 Rolf Bader: Multiple vihuela sound hole radiation 821

Downloaded 12 May 2013 to 131.104.62.10. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
very fast. The point of minimum energy is therefore also middle between point j and the adjacent point to the struc-
intuitively the correct one and we can very reasonably follow ture, then l ¼ g/2 and therefore wj ¼ 0.5.
the reconstruction visually by watching the solution getting Additionally, for comparing the phases of the sound
sharper with decreasing a up to the optimum point. holes for the Herb vihuela, its complex sound hole radiation
Two different grid constants g were used to meet the dif- was calculated, too, like
ferent instrument sizes. With the LK classical guitar and the ð
Herb vihuela g ¼ 4.75 cm and with the Yaish vihuela g ¼ 3.9 ~
Si ðf Þ ¼ Aj ðf Þ only for Herb with
cm were used. All six open strings were plucked and recorded Area Sound holeðsÞ

at each instrument, where the tuning of the classical guitar i ¼ 1; 2; 3; 4; 5: (10)

was taken to be at modern standard pitch of 440 Hz for a1 and
the tunings of the vihuelas to be at historical 415 Hz for this To have a value for the phase interactions between the sound
pitch. As vihuelas are normally tuned like lutes with strings holes, for three configurations, the normalized absolute value
tuned to G-c-f-a-d0 -g0 (modern classical guitar tuning is E-A- of the complex sum of the sound hole radiation was calcu-
d-g-b-e0 ), the lowest pitch of the vihuela is at 90 Hz, while the lated for all sound holes like
lowest pitch of the classical guitar is at 83 Hz. X 
 S~i ðf Þ
The resulting complex amplitude radiation patterns
i
A~j ðf Þ for microphone j ¼ 1,2,3,…,121 and frequency f were SHerb
all ðf Þ ¼
X ; i ¼ 1; 2; 3; 4; 5; (11)
normalized for each frequency. Then its absolute value Si ðf Þ
i
Aj ðf Þ ¼ jA~j ðf Þj was calculated and then split into the follow-
ing parts: for the left/right split (only using the larger, lower sound
holes) like
(a) sound hole(s) Si(f) (i ¼ 1 for LK and Yaish, i ¼ 1,2,3,4,5 X 
for Herb),  S~i ðf Þ
i
(b) bridge B(f), SHerb
lf ðf Þ ¼ X ; i ¼ 2; 3; (12)
(c) and top plate T(f) without sound hole(s) and bridge, inte- Si ðf Þ
grating the amplitudes over the areas of the sound holes i

like and for the up/down split (only using the left sound holes as
ð they are radiating stronger all over) like
Sðf Þ ¼ Aj ðf Þ; (6) X 
Area Sound holeðsÞ  S~i ðf Þ
i
SHerb
ud ðf Þ ¼
X ; i ¼ 2; 4: (13)
the bridge like Si ðf Þ
ð i

Bðf Þ ¼ Aj ðf Þ; (7) The numbering of the Herb sound holes is 1 ¼ middle, 2 ¼ lower
Area Bridge
left, 3 ¼ lower right, 4 ¼ upper left, and 5 ¼ upper right.
and the top plate like In the calculations of the Helmhholtz resonance Eq. (2)
for the vihuelas and Eq. (1) for only the Herb vihuela, the
ð reduced surfaces were taken into account. Therefore the
Tðf Þ ¼ Aj ðf Þ: (8) reduced radius was calculated to be
Total AreaArea BridgeArea Sound holeðsÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffi
rred ¼ 0:65r 2 : (14)
Here the area integrals were performed as weighted sum
over the reconstruction points j, like IV. RESULTS
ð X A. Helmholtz resonances
Aj ðf Þ ¼ ¼ wj A~j (9)
Area j The Helmholtz resonances of the three instruments are
(a) Yaish vihuela, 130 Hz,
where all reconstruction amplitudes laying on the respective
(b) Herb vihuela, 138 Hz,
structures (holes, bridge, top plate) have wj ¼ 1, all points
(c) LK guitar 98 Hz.
next to a structure have wj ¼ l/g (so using a linear interpola-
tion) and all points at least one point j away have wj ¼ 0. They were determined by knocking on the bridges,
This interpolation is necessary if the structure ends between while the instruments were still mounted in front of the
two grid points. In this case, l is the distance between the last arrays. From the radiating patterns the Helmholtz frequen-
grid point j on the structure and the end of the structure. So cies could clearly be associated to spectral peaks. Figures
to take the radiation of the structure beyond the last grid 3(a) to 3(c) show the spectra of (a) the Yaish vihuela, (b) the
point into consideration, the radiation of the nearest grid Herb vihuela, and (c) the LK guitar with peaks at the Helm-
point next to the structure is added to the integral with the holtz resonances at 130, 138, and 98 Hz, respectively, which
weight wj. So, e.g., if the end of the structure is right in the are always the lowest peak in the spectra. At the bottom of

822 J. Acoust. Soc. Am., Vol. 131, No. 1, Pt. 2, January 2012 Rolf Bader: Multiple vihuela sound hole radiation

Downloaded 12 May 2013 to 131.104.62.10. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
FIG. 3. Spectra (top) and radiation patterns of the Helmholtz resonance peaks (bottom) for (a) the Yaish vihuela, (b) the Herb vihuela, and (c) the LK guitar.
Each spectrum is normalized with respect to its largest peak.

the three plots, the respective radiation patterns at these expected. Still it is interesting to see that the much smaller
frequencies on the instrument top plates as measured and body of the Yaish vihuela has an even lower Helmholtz reso-
back-propagated by the microphone array are displayed. nance. On the other hand, both instruments are tuned to the
Both single sound hole instruments clearly show a radiation same pitches, and therefore the instrument builder need to
restricted to the sound hole. With the Herb vihuela, the radi- meet a lower Helmholtz resonance frequency to make the
ation at the sound holes 2 and 3 is strongest while sound bass notes loud enough. Also the lowest note of the classical
holes 1, 4, and 5 show much less radiation. This strong dif- guitar of 83 Hz is a bit lower than that of the vihuela of about
ference in radiation strength depends mostly on the sound 90 Hz (depending on the tuning system used) and so again, a
hole positions and is discussed in the following text. When higher Helmholtz is needed.
comparing the phases of these sound hole radiations to the Table I shows measured and calculated frequencies for
phase of the middle hole 1 (according to the Herb sound all three instruments using the single sound hole Eq. (1) and
hole numbering from above) as fractions of 2p like additionally the multi sound hole Eq. (2) for the Herb
d/n ¼ (/1  /n)/(2p), we get d/2 ¼ 0.0096, d/3 ¼ 0.0108,
d/4 ¼ 0.0290, and d/5 ¼ 0.0122. Therefore we may argue
that the radiation from the five holes are all in-phase as TABLE I. Measured and calculated Helmholtz resonances for the three
expected from a Helmholtz resonance. So with all instru- instruments using the single sound hole Eq. (2) with two variations (radia-
tion relation, reduced radii) and the multi sound hole Eq. (1) (only for Herb
ments the Helmholtz resonance peaks in the spectrum could vihuela).
clearly be identified as such by examining the radiation pat-
terns at these peaks. Yaish vihuela Herb vihuela LK guitar
Further associations between peaks and radiation pat-
Measured 130 138 98
terns were also possible. These patterns are often quite com-
Eq. (2) full radii 121 145 96
plex especially with the Herb vihuela where new patterns Eq. (2) radiation relation – 137 –
arose. Going into details here is beyond the scope of this Eq. (2) reduced radii 109 130 –
paper, which concentrates on the sound hole radiation. Eq. (1) full radii – 121 –
As the bodies of the vihuelas are smaller than that of a Eq. (1) reduced radii – 104 –
classical guitar a raise of the Helmholtz resonance was

J. Acoust. Soc. Am., Vol. 131, No. 1, Pt. 2, January 2012 Rolf Bader: Multiple vihuela sound hole radiation 823

Downloaded 12 May 2013 to 131.104.62.10. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
vihuela. As the rosettas of the vihuelas indeed have quite the surface areas and again calculating the Helmholtz for the
large surfaces where no air flow can happen, the reduced Herb vihuela, we get 137 Hz being very close to the meas-
radii rred were used in these cases too. This was omitted for ured 138 Hz. The reason for the different radiation strength
the classical guitar as it has no rosetta. Here the original ra- of the sound holes is discussed in the following. It reflects
dius still holds. the amount of participation of the sound holes to the Helm-
For the LK guitar, the calculated frequency of 96 Hz holtz resonance.
meets the real one of 98 Hz very well as expected pointing
to the stability of the calculation for guitar sound holes. With
B. Radiation strength of sound hole(s) compared to
the Yaish vihuela the situation is much different. Its calcu-
bridge and top plate
lated frequency of 109 Hz is 21 Hz below the measured one
of 130 Hz. The frequency raises if the original radius of The three instruments clearly show different behavior
r ¼ 3.55 cm is used instead of the reduced one rm ¼ 2.86 cm concerning the radiation strength of the sound holes. In Fig.
up to 122 Hz. 4, the frequency dependent radiations are shown for all 60
As the radius reduction equation fails much more than pitches used with each instrument, where the lowest is at 83
the full sound hole equation, in a second experiment, the Hz (90 Hz) and the highest is about 3 kHz. Here the sound
influence of the rosetta onto the Helmholtz resonance was hole S(f), the bridge B(f), and the top plate T(f) radiation
tested. Here the LK guitar was used for not to destroy the strengths as sums of absolute radiations are shown. The
vihuela rosettas. Its Helmholtz was again measured by absolute values are taken as an objective value, as a sum of
recording a knocking on the bridge while covering its sound the complex amplitudes would depend upon the choice of a
hole. First, the hole was covered by a metal fence taken from listening or recording position in the space in front of the
the protection of a loudspeaker fitting perfectly into the LK instrument.
sound hole covering about half of its surface area. Surprisingly, Still all three have a different character. The LK guitar
the Helmholtz increased with this fence, although only slightly at the bottom of Fig. 4 has the highest sound hole radiation
by 1.1 Hz. Second, a thin plastic layer was used to cover the peak a bit above its Helmholtz resonance as expected. Still
sound hole. Again its Helmholtz raised only by 0.6 Hz. with a maximum of about 70% of the hole instrument radia-
The mounting of the fence is not expected to affect the top tion at this point, it is by far the strongest sound hole radia-
plate as it could solely be fixed at the fretboard of the guitar tion of all instruments. But this strong radiation is limited to
that ends right at the sound hole. Still, the plastic layer was about 300 Hz. Then a region appears between about 200 and
fixed at the sound hole rim and could have a slight effect. The 400 Hz, where the bridge has its radiation peak. Above this
raising of the frequency points to a increase of the end- region, the sound hole shows another strong radiation
correction factor .85 in Eq. (2) rather than to a reduction of the around 400 Hz, only to decay again. From 500 Hz on, the
sound hole radius which would have lowered the Helmholtz radiation of the guitar is clearly dominated by the rest of the
resonance. top plate.
So we are allowed to use the original radius for the The Yaish vihuela has an interesting parallel behavior
Yaish vihuela ending at a Helmholtz of 122 Hz with Eq. (2). compared to the LK guitar. The sound hole again is strong
This frequency could be raised by the rosetta was a result of in the bass region with a peak around its Helmholtz reso-
a change in the end-correction value, but this raise would be nance of 130 Hz as expected. Its maximum peak of about
expected only to be around 1 Hz. Although still a bit flat, the 50% of the total radiation is not as strong as with the LK.
calculated Helmholtz is still reasonable near to the measured Indeed the Yaish is not that strong in the bass, which again
one. is expected as its body is much smaller. It is also interesting
The Herb vihuela deviated from the measured Helm- to see that the frequency region in which the sound hole is
holtz frequency by 8 Hz from Eq. (2) and by 29 Hz from Eq. strong is again limited up to about 300 Hz as is the case
(1). Here the full radii version of Eq. (2) gives a frequency with the LK guitar. Again parallel to the LK, the Yaish also
too high, and the reduced radii version gives a frequency too shows the strongest bridge radiation region around 400 Hz.
low. We have seen with the experiment of the fence with the This similarity seems to follow the idea of construction for
LK guitar discussed in the preceding text that the radius “modern” vihuelas, namely to have a mixed sound between
reduction does not lead to a lowering of the Helmholtz. Still lute and guitar. Basically a guitar with similar pitch region
when examining Fig. 3(b), with the radiation pattern of the and built with ribs has a bass region too weak to come close
Herb vihuela Helmholtz clearly some sound holes radiate in sound to a lute. All in all, compared again with the LK,
much stronger than others. Therefore weights for the sound the vihuela shows a much smoother transition from the bass
holes are introduced as fractions of their radiation strength region to the trebles in terms of radiation strength of its
like parts.
sffiffiffiffiffiffiffiffiffiffi The Herb vihuela with five sound holes again show a
jS~i j different behavior. The sound hole has a much larger fre-
bi ¼ ; (15) quency range with strong radiation up to around 500 Hz. The
jS~max
highest peak is again about 50% of the total radiation at its
where again the numbering of the sound holes i ¼ 1,2,3,4,5 Helmholtz resonance of 138 Hz. Here the bridge has not
for the Herb vihuela is used. jS~max j is the amplitude of the such a clear strongest region as with the LK and the Yaish
strongest sound holes. When multiplying these weights with instruments.

824 J. Acoust. Soc. Am., Vol. 131, No. 1, Pt. 2, January 2012 Rolf Bader: Multiple vihuela sound hole radiation

Downloaded 12 May 2013 to 131.104.62.10. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
 sound holes are perfectly in-phase, the value is one. If they
Herb
have an anti-phase relation of p, this value is zero. S all is
Herb
summing over all sound holes, Slr is looking for the left/
right dependency taking only the lower (larger) sound holes
into consideration as they have a stronger radiation power.
Herb
Sud sums the two left sound holes (again radiation stronger
than the right ones), so comparing the lower and higher
sound holes. The reason for this may be seen in differences
in the top plate construction, which cannot be crafted per-
fectly symmetric.
In Fig. 5 it clearly appears that the phase relations up to
about 500 Hz of the sound holes are more or less all in-
phase. This is most strongly the case for the left/right split.
The upper/lower split shows a slightly lower phase consis-
tency pointing to a slight phase difference between these two
holes. The behavior is plausible within the low frequency
range. Here long wavelength appears in the sound box
mostly along its long side. So the left/right phases need to be
the same. Still along the up/down direction of the body, the
phase will change constantly, and when taking the phase at
two arbitrary points, it would be necessary for them to show
a slight phase difference like different points on a wave
show. The phase behavior of all five sound holes being
placed in between these two is reasonable then as a medium
value of phase consistency between right/left and up/down
directions.

D. Radiation strength of the five sound holes of the

Herb vihuela
The values of Si(f), the absolute radiation strength of the
five sound holes of the Herb vihuela, show a clear picture,
too, as shown in Fig. 6. The strength of the radiation depends
upon the position of the hole, its distance to the mid point of
the sound box. The two lower holes on the left and right
clearly show the strongest radiation. The upper left and right
holes have a quite low radiation. The middle hole positioned
not only in the middle of the vihuela axis but also in the mid-
dle in terms of its height is indeed medium in radiation

FIG. 4. Frequency dependent radiation areas (normalized absolute radiation

amplitudes) for the three instruments used; top: Herb vihuela (5 sound
holes), middle: Yaish vihuela (1 sound hole), bottom: LK classical guitar.
The radiation area of the sound hole(s), the bridge, and the rest of the top
plate are shown. They are calculated by integrating the radiation patterns of
the fundamental frequency and nine of its overtones (or equivalently the first
10 partials) of six open strings plucked on each instrument. So the frequency
resolution depends upon these overtones and is in the low region around 30
Hz. With the LK, the sound hole dominates the radiation in the bass register
up to about 300 Hz. The Yaish vihuela shows about the same appearance of
the sound hole radiation but less strong than the LK. The Herb vihuela has a
much broader frequency region of the sound hole radiation up to about
500 Hz.

C. Phase relations of the five sound holes of the Herb

vihuela
FIG. 5. Phase relations between the five sound holes of the Herb vihuela
Taking a closer look at the sound hole behavior in Fig. depending on frequency calculated as normalized complex amplitude sum
for all five sound holes, for lower left and right holes and left upper and
5, the phase relations of the sound holes are shown using the
lower holes. Likewise the radiation area showing an area drop around 500
normalized absolute values of the sum here of the complex Hz (see Fig. 2, bottom plot) phase differences cancel out different holes
radiation amplitudes SHerb (see Sec. III). If the phases of the leading to reduced radiation strength from about 500 Hz on.

J. Acoust. Soc. Am., Vol. 131, No. 1, Pt. 2, January 2012 Rolf Bader: Multiple vihuela sound hole radiation 825

Downloaded 12 May 2013 to 131.104.62.10. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
 where lx , ly , lz are the length, width, and height of the box.
As the sound holes of the Herb instrument start at about half
of its length, the mode of the lowest frequency causing
sound holes to radiate in anti-phase is (2,0,0), where one
nodal line would lie somewhere through the sound holes,
with a frequency of f (2,0,0) ¼ 715 Hz. Still, as the phase dif-
ferences start at about 500 Hz, we would need to add to the
length lx a term to take the influence of the sound holes into
consideration. As this mode has no movement in the top
plate or back direction, only the ribs would add some relaxa-
tion to the vibration and therefore would lead to a frequency
reduction. As is known from guitars, low vibrational modes
are all combinational modes where the air couples to the gui-
tar top and back plates and vice versa. Similarly, a standing
FIG. 6. Absolute amplitude strength for the different sound holes of the wave in the ribs direction will couple to the ribs leading to a
Herb vihuela. The lower holes which are larger in diameter have the highest combinational mode there, too. Still this effect is not
radiation strength (lower left, lower right). The smaller and upper holes
have the lowest radiation (upper left, upper right). The mid hole (which is as expected to be so strong as to lead to a frequency decrease of
large as the lower holes) is in between these radiations pointing to a position more than 200 Hz with quite stiff ribs. If we add a length
dependency of the holes in terms of radiation strength. Also a tendency for correction d for the vihuela length lx like
the left holes to radiate stronger than the right ones appears.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
c
f ðn; m; lÞ ¼ ðn=ðdlx ÞÞ2 þ ðm=ly Þ2 þ ðl=lz Þ2 ;
strength compared to the upper and lower sound holes. From 2 (17)
these data it is hard to say if the fact that the upper holes are m; n; l ¼ 0; 1; 2; 3;:::;
smaller than the lower ones is causing this effect or the lower
air movements of holes closer to the ribs. Still, as the middle then, assuming f (2,0,0)  500 Hz, we can calculate d  1:43
hole, which has the same size as the lower ones is closer to taking mainly the sound hole and to a slight extend the ribs
these upper holes in terms of radiation strength than to the influence on the (2,0,0) mode into account. This is reasona-
lower ones, it may be assumed that the position of the sound ble as then we have f (1,0,0)  350 Hz using the equation
hole in the instrument is determining its radiation strength above. Indeed, this clearly appears as resonance peak in
much stronger than its actual size (if its size is kept within a Fig. 6.
reasonable range). Therefore the reason for the wide frequency range of
Also the left holes show a stronger radiation than the strong sound hole radiation with the Herb vihuela may be
right ones. As this vihuela is built perfectly symmetric, no explained as follows. Normally, the Helmholtz resonance is
geometrical reason can be found for this behavior, it must be radiating much stronger through the sound hole than higher
connected to the pitches played on the instrument. As the air modes do. This is not the case for the five hole vihuela
bass strings are attached to the left side and drive the bass as the sound hole area is much larger here. This large area
frequency region, this enhanced left-sided radiation may be lead to an overall high sound hole radiation as long as all
caused by the driving point. Still further detailed studies sound holes act like one monopole, which they do only up to
would be needed to make this behavior clear. 500 Hz.
The classical guitar shows a very prominent sound hole
E. Reason for different sound hole radiation behavior radiation, making it much stronger in the bass region than
the vihuelas. Still when higher nodes appear, these modes do
Taking the phase relations, the sound hole radiation radiate much less as the radiation area is also much less
strengths, and the Helmholtz resonance frequencies into con- compared to the five tone holes of the vihuela. Also the
sideration, an interesting behavior of higher order air har- Yaish vihuela has only one sound hole and so comes closer
monics in the multi sound hole vihuela can be assumed. As to the classical guitar. So the sound hole radiation frequency
can be seen in Fig. 5, the phase relation of the up/down range of this vihuela has about the size of the classical guitar,
sound holes is considerable faster getting out of phase than still it is not that strong simply as the instrument body is
the left/right one from 500 Hz on. This means that the lower smaller.
holes start vibrating in anti-phase with the higher ones for
higher frequencies. The waist of the vihuela is not so promi-
nent, and we can therefore approximate the vihuela body as F. Examples of radiation patterns
a box by averaging over the body curves. Then the air Two examples of radiation patterns are shown in Fig. 7
enclosed as an air volume for low frequencies can be for the Yaish vihuela and in Fig. 8 for the Herb instrument as
approximated as standing waves with frequencies of examples for radiation patterns of interesting frequency
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi regions.
c The fifth partial of the low G-string played on the small
f ðn; m; lÞ ¼ ðn=lx Þ2 þ ðm=ly Þ2 þ ðl=lz Þ2 ;
2 (16) Yaish vihuela has a frequency of 463 Hz and so is an exam-
m; n; l ¼ 0; 1; 2; 3;:::; ple of the interesting frequency region around 500 Hz, where

826 J. Acoust. Soc. Am., Vol. 131, No. 1, Pt. 2, January 2012 Rolf Bader: Multiple vihuela sound hole radiation

Downloaded 12 May 2013 to 131.104.62.10. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
FIG. 7. Example of a radiation pattern of the Yaish vihuela where the sound FIG. 8. Example for a radiation pattern of the Herb vihuela where four
hole, the bridge, and the top plate are all radiating strongly, which is the sound holes, the bridge, and the top plate are all radiating strongly for the
case for this instrument around 500 Hz (see Fig. 2, middle plot). Here the lowest pitch on the instrument (lowest G-string, 1st partial at 90 Hz). As the
low G-string of its 5th partial at 463 Hz is taken as an example for a pattern Helmholtz resonance is at 138 Hz with this instrument, below this frequency
around this frequency. the strength of the top plate is strengthened again.

the bridge shows its strongest radiation while the sound hole higher amount of holes lead to a much larger region of sound
and the top plate are radiating, too. It can be seen that the hole radiation. As the single hole instruments drop the main
bridge radiation is connected to a strong vibration at the right radiation strength at around 300 Hz, the five hole instrument
side of the instrument next to the bridge. The radiation of the does so at around 500 Hz. Below this the phase relations
sound hole also shows some movement of the top plate between the sound holes are very phase consistent one to
around it. This is pointing to the top plate region around the another and so cause only very low cancellation between the
hole being more flexible than other top plate regions closer holes. Above 500 Hz, the (2,0,0) mode is the first to have a
to the ribs, which are much stiffer. nodal line right through the sound holes causing anti-phase
Figure 8 shows the radiation pattern of the Herb vihuela relations between the sound hole radiations and therefore a
for the lowest pitch. This radiation is very interesting as not considerable radiation drop. The larger frequency region of
only the sound holes radiate quite strongly but also the bridge, strong sound hole radiation with the five sound hole instru-
and the the top plate next to the bridge on the right and at the ment is mainly caused by the larger radiation area of these
very low end of the instrument also show strong radiation. sound holes.
This seems to be the case because the frequency of this pattern It also appears that the position of the holes on the top
of 90 Hz is quite a lot below the Helmholtz resonance of the plate leads to very different radiation strengths, where sound
instrument of 138 Hz (48 Hz lower). Within the Helmholtz holes closer to the middle of the sound box radiate much
frequency region, the radiation pattern shows much less radia- stronger than those at the sides. Vihuelas seem to have a
tion from the bridge or top plate than it does within the lower higher Helmholtz resonance, where instrument builders try
frequency region. Classical guitars do not show such behavior to fit the size of the sound holes and the enclosed air volume
as the Helmholtz resonance around 100 Hz is closer to the in a way to tune the Helmholtz resonance frequency to about
lowest pitch of 83 Hz (17 Hz lower) than it appear with vihue- 130 Hz. This is clearly meant as the two vihuelas differ in
las. This is the cause for the drop of the sound hole radiation size quite a lot. Here the radiation strength relations between
shown in Fig. 4, bottom plot, below 100 Hz. the sound holes change the Helmholtz resonance frequency.
It also appears that the role of the sound hole radiation
changes over frequency much more abruptly with the classi-
V. CONCLUSION
cal guitar compared to the vihuelas, which fade much more
The amount and placing of sound holes play an impor- smooth between these two regions. Still the guitar shows
tant role in the sound radiation of vihuelas. Comparing the much stronger radiation of the sound hole within this low
single hole vihuela and guitar to a five hole vihuela the frequency range compared to the vihuelas.

J. Acoust. Soc. Am., Vol. 131, No. 1, Pt. 2, January 2012 Rolf Bader: Multiple vihuela sound hole radiation 827

Downloaded 12 May 2013 to 131.104.62.10. Redistribution subject to ASA license or copyright; see http://asadl.org/terms
 4
The vihuelas used here are not historical instruments of A. Le Pichon, S. Berge, and A. Chaigne, “Comparison between experi-
which only three have survived and are not playable. In addi- mental and predicted radiation of a guitar,” Acust. Acta Acust. 84(1),
136–145 (1998).
tion, customers’ needs were considered in the instruments’ 5
A. Peekna and Th. D. Rossing, “The acoustics of carved Baltic psalteries,”
construction, making the sound of the historic instrument Acust. Acta Acust. 91(2), 269–276 (2005).
6
very hard to reconstruct. When interest in historical perform- A. Peekna and Th. D. Rossing, “Tuning the lower resonances of carved
Baltic psalteries by adjusting the areas of the sound holes,” J. Acoust. Soc.
ance practice started about 30 years ago, builders were asked
Am. 117(4), 2590 (2005).
to build vihuelas, which were only known from historical 7
N. Fletcher and Th. D. Rossing, Physics of Musical Instruments (Springer,
paintings and a few historical ones. This led builders to New York, 2000), p. 14.
8
guess that the sound of a vihuela was between a lute and a M. McPherson, “Arrangement of a sound hole and construction of a sound
board in an acoustic guitar,” J. Acoust. Soc. Am. 117(5), 2699 (2004).
guitar. Indeed, all kinds of vihuelas existed, from soprano 9
W. Y. Strong, Jr., T. B. Beyer, D. J. Bowen, E. G. Williams, and J. D.
ones to contra-bass instruments, and they may have sounded Maynard, “Studying a guitar’s radiation properties with nearfield hol-
very differently. From this standpoint, the investigated ography,” J. Guitar Acoust. 6, 50–59 (1982).
10
instruments cannot be taken as “real historic” but as interest- G. Weinreich, “Sound hole sum rule and the dipole moment of the violin,”
J. Am. Acoust. Soc. 77(2), 710–718 (1985).
ing examples of a hybrid lute and guitar instruments and 11
G. Bissinger, E. G. Williams, and N. Valdivia, “Violin f-hole constribution
examples of multi sound hole radiators. to far-field radiation via patch near-field acoustical holography,” J. Acoust.
Soc. Am. 121(6), 3899–3906 (2007).
12
R. Bader, “Reconstruction of radiating sound fields using minimum energy
ACKNOWLEDGMENTS method,” J. Acoust. Soc. Am. 127(1), 300–308 (2010).
13
M. Ochmann, “The full-field equations for acoustic radiation and
I thank my colleague Olaf Van Gonnissen for supplying scattering,” J. Acoust.Soc.Am. 105(5), 2574–2584 (1999).
14
me with the Herb vihuela. I also thank the guitar workshop E. G. Williams, Fourier Acoustics. Sound Radiation and Nearfield Acous-
of Alexander and Hermann Herb for giving me valuable tic Holography (Academic, San Diego, 1999).
15
S. F. Wu, “On reconstruction of acoustic pressure fields using the Helm-
details about the vihuela from their workshop. holtz equation least squares method,” J. Acoust. Soc. Am. 107(5),
2511–2522 (1999).
1 16
O. Schöner, Die Vihuela de Mano im Spanien des 16. Jahrhunderts (The M. B. S. Magalhães and R. A. Tenenbaum, “Sound source reconstruction
Vihuela de Mano in 16th Century Spain) (Verlag, Frankfurt am Main, 1999). techniques: a review of their evolution and new trends,” Acust. Acta
2
I. Woodfield, The Early History of the Viol (Cambridge University Press, Acust. 90, 199–220 (2004).
17
Cambridge, UK, 1988). M. J. Elejabarieta, C. Santamaria, and A. Ezcurra, “Air cavity modes in
3
For illustrations see http://www.lutesandguitars.co.uk/htm/cat12.htm (Last the resonance box of a guitar: The effect of the sound hole,” J. Sound
viewed October 2010). Vibr. 252(3), 584–590 (2002).

828 J. Acoust. Soc. Am., Vol. 131, No. 1, Pt. 2, January 2012 Rolf Bader: Multiple vihuela sound hole radiation

Downloaded 12 May 2013 to 131.104.62.10. Redistribution subject to ASA license or copyright; see http://asadl.org/terms