Sie sind auf Seite 1von 15

Applied Acoustics 56 (1999) 183±197

Tuning characteristics, radiation eciency and


subjective quality of a set of classical guitars
Ricardo R. Boullosa *, Felipe OrdunÄa-Bustamante 1,
A. PeÂrez LoÂpez 2
SeccioÂn de AcuÂstica, Centro de Instrumentos UNAM, Circuito Exterior CU, Apdo. Postal 70-186, CP 04510,
MeÂxico DF, MeÂxico

Received 1 June 1997; accepted 1 April 1998

Abstract

This paper is an exploratory study of the in¯uence of tuning and radiation properties on the
subjective quality of a small set of classical guitars. It is shown that overall measures of the
tuning errors and of the radiation eciency below 1 kHz produce good estimates of the sub-
jective musical quality of the guitars. # 1998 Elsevier Science Ltd. All rights reserved.
Keywords: Subjective quality; Radiation eciency; Guitar acoustics.

1. Introduction

Research on the subjective quality of classical guitars [1±4] has revealed some
correlation with physical characteristics such as atack and decay times of the tones,
sound pressure level, natural frequency and bandwith of the lower guitar reso-
nances, etc. But, as far as we know, this is the ®rst attempt trying to correlate tuning
properties and radiation eciency of the guitars with their subjective musical qual-
ity. We believe these properties can be more intimately related to the cues subjects
use to rate the sound quality of the instrument. For one thing, the tuning properties
are a measure of how accurately the guitar plays (both melody and harmony, and
also how well the guitar ®ts in a musical ensemble), and radiation eciency is related
to how loud the guitar sounds. But for another, both properties are also related to
how easy it is for the performer to produce the sound he/she likes, something which
ultimately determines the quality of his/her performance.
*Corresponding author. E-mail: rrb@aleph.cinstrum.unam.mx;
hhtp://aleph.cinstrum.unam.mx/acustica.html
1
E-mail: felipe@aleph.cinstrum.unam.mx
2
E-mail: toni@aleph.cinstrum.unam.mx

0003-682X/99/$Ðsee front matter # 1998 Elsevier Science Ltd. All rights reserved.
PII: S000 3-682X(98)0002 7-9
184 R.R. Boullosa et al./Applied Acoustics 56 (1999) 183±197

More than producing strong experimental evidence, the aim of this paper is to
suggest new ways to try on the search for physical correlates of the subjective quality
of the classical guitar. After giving a brief description of the guitars involved in this
study, sections of the paper are devoted to the tuning properties of the guitars, the
measurement of the radiation eciency and the tests to determine subjective quality.
At the end some conclusions are given.

2. The guitars

The guitars involved in this study range from `very-low' to `medium-high' quality
(as judged by the authors and by the people who played the guitars during this
study). A more objective statement is, perhaps, to say that their prices range from
about US $50 to $500. The guitars have been given three-letter code names (which in
three instances actually correspond to the initials of their owners). A brief (sub-
jective) description of each guitar follows.

CHF: Made by an unknown maker in Mexico City. A `very-low' quality guitar (in
the $50 US dollar range) which is hard to play. Its timbre is not particularly nice and
its loudness is low.3
APL: Made by an unknown maker in Paracho, Mexico. A medium quality guitar
(probably in the US $100 range) which is moderately easy to play. It has a ®ne
timbre but its sound is not particularly loud.
RRB: Yamato model Y93000 made in Mexico. A medium quality guitar (probably in
the US $150±300 range), not very easy to play (the action being perhaps a bit high, i.e. the
string slightly hard to press). It has a regular timbre and has a moderately loud sound.
FOB: Model 4P made by `Alhambra' in Spain. A `medium-high' quality guitar (in
the US $500 range), judged to be very easy to play (with a medium to low action). It
has a ®ne timbre and a moderately loud sound.

All the guitars are in good condition and each was ®tted for this study with the
same brand of nylon and metal-wrapped strings (La Bella, 900B Golden Superior,
Classical Guitar Strings).

3. Tuning characteristics of the guitars

The actual pitch of the notes played on a string stopped at di€erent frets can
depart from the exact nominal pitch (assigned to each fret) depending on several
factors. The following can be mentioned as the most important:

3
The three-letter code for this guitar does not correspond to the initials of its owner, who has gra-
ciously remained anonymous (actually, the people in our lab do not know, or pretend not to know, who
owns this guitar). The code-name CHF stands for `chafa', which in Mexican slang denotes something of a
rather poor quality.
R.R. Boullosa et al./Applied Acoustics 56 (1999) 183±197 185

. The exact position of the frets along the fretboard.


. The distance between the string and the fret (and how this distance varies along
the fretboard).
. The coupling of the mechanical oscillations of the string and those from the
body of the guitar [5].

Variations in pitch induced by these (and possibly other) factors de®ne the tuning
characteristics of the guitar.
Pitch is a subjective attribute of sound and it is generally dicult to trace pitch
variations back to their physical causes. This traceability is important if we want to
relate the subjective musical quality of the guitar to the physical characteristics of
the instrument (and, hopefully, give some insight into how to modify the instrument
to improve its quality). The experimental results quoted below use the fundamental
frequency of vibration of the string as an indication of pitch. This is an inaccurate
indication because pitch information contained in the higher partials of the string
and in the temporal evolution of the tone are completely left out. However, we trust
some information about pitch is still contained in the fundamental frequency of the
string and, based on this assumption, we exploit the fact that variations in the fun-
damental frequency can be easily traced back to their physical causes.
The fundamental frequency of vibration of the strings, open and stopped at each
fret, was measured in the four guitars involved in this study. The procedure was as
follows. The fundamental frequency was monitored with a spectrum analyzer B&K
2034, the sound being picked-up by a microphone B&K 4134. Working with one
string at a time, the fundamental frequency of the open string was measured and
adjusted to its nominal tuning. Next, the fundamental frequency of the string stop-
ped at each fret was measured. During this process, periodic checks were made of
the tuning of the open string. The deviation of the measured tuning (i.e. frequency)
fm with respect to the nominal tuning fn was expressed in cents using the formula

T ˆ 1200 log2 …fm =fn † …1†

where 100 cents=1 semitone and 1200 cents=1 octave. Fig. 1(a) shows the tuning
characteristics measured in this way in the guitar code-named CHF. Note that fret 0
indicates the measured tuning of the open string. Tuning data were obtained up to
fret 19, but only 12 frets are shown for clarity.
The mechanical response of the guitars was also measured. For this, the mechan-
ical inertance (the complex ratio between acceleration and force) was measured
using the same spectral analyzer B&K 2034. An impedance head B&K 8001 (a force
and acceleration transducer) was used to excite the bridge of the guitars at the point
of rest of the third string. The impedance head was attached to and driven by a
vibration exciter B&K 4809 fed with a random signal. A combined graph of the
tuning characteristics and the (uncalibrated) log-magnitude of the mechanical
response as a function of frequency is shown in Fig. 1(b). This ®gure shows quite
clearly the e€ect of string-body coupling on the tuning of the guitar, particularly
that of the two lower resonances of the guitar on the tuning of strings 4, 5 and 6.
186 R.R. Boullosa et al./Applied Acoustics 56 (1999) 183±197

Fig. 1. Tuning shift for strings 1 to 6 in guitar CHF (a) As a function of fret and (b) as a function of
frequency. A footprint of the mechanical response of the guitar is shown at the bottom.

Fig. 2. Tuning shift for strings 1 to 6 in guitar APL (a) as a function of fret and (b) as a function of
frequency. A footprint of the mechanical response of the guitar is shown at the bottom.
R.R. Boullosa et al./Applied Acoustics 56 (1999) 183±197 187

Fig. 3. Tuning shift for string 1 to 6 in guitar RRB (a) as a function of fret and (b) as a function of fre-
quency. A footprint of the mechanical response of the guitar is shown at the bottom.

Fig. 4. Turning shift for string 1 to 6 in guitar FOB (a) as a function of fret and (b) as a function of
frequency. A footprint of the mechanical response of the guitar is shown at the bottom.
188 R.R. Boullosa et al./Applied Acoustics 56 (1999) 183±197

The tuning characteristics for the other guitars are shown in the corresponding
®gures and illustrate other aspects of the tuning problem. Guitar APL shows a
consistent sharpening of the notes suggesting that string±fret separation increases
along the fretboard. Guitar RRB shows a sudden sharpening of the notes as soon as
the strings are stopped from the ®rst fret upwards but showing no further sharpen-
ing. This suggests that this guitar has an unusually large string±fret separation at the
position of the ®rst fret which causes the observed behaviour. Guitar FOB has very
similar tuning characteristics to guitar CHF. All the guitars show a more or less
pronounced e€ect of string±body coupling on tuning, but only guitars APL and
RRB show e€ects from string±fret separation. There is apparently no e€ect of fret
position on the tuning of these guitars.
In order to integrate the tuning properties of the guitars into simple ®gures, the
mean and maximum tuning errors were calculated for each guitar (using data for
strings 1 to 6 and frets 1 to 19). In addition to raw tuning errors, the fret-to-fret
interval error was also analysed. The results are shown in Figs. 5 and 6. Fig. 5 shows
that the mean tuning error varies noticeably among the guitars (ranging from 7 cents
to 20 cents in guitars FOB and RRB respectively), whereas the mean interval error
(at about 6 cents) is remarkably similar in all the guitars. Fig. 6 shows that the
maximum tuning error and the maximum interval error both vary among the guitars
(taking values in the range from 20 to 50 cents). However, the relative values are
distributed di€erently among the guitars, so that a list of the guitars sorted by max
tuning error (CHF, APL, FOB, RRB) is di€erent from a list sorted by max interval
error (APL, CHF, RRB, FOB) and both lists are yet di€erent from a list sorted by
mean tuning error (FOB, CHF, APL, RRB). The question of which of these tuning

Fig. 5. (a)Mean tuning error and (b) mean interval error.


R.R. Boullosa et al./Applied Acoustics 56 (1999) 183±197 189

Fig. 6. (a)Max tuning error and (b) max interval error.

®gures is more important with respect to the subjective quality of the guitars will be
discussed later in the conclusions.

4. Radiation eciency

The radiation eciency (the ratio of radiated sound power to the input mechan-
ical power) of each guitar, with muted tuned strings, in one-third octave bands from
100 Hz to 3150 Hz, was determined following the same general procedure and for-
mulation given by Susuki [6,7]. The basic di€erence being that instead of measuring
sound intensity, the radiated sound power in a reverberant chamber of 200 m3 (using
the standard method described in ISO-3742 and ANSI-S1.32) was determined
simultaneously with the input power, obtained with a two-channel analyzer B&K
2034. The input power can be obtained from the imaginary part of the cross-power
spectrum between acceleration (divided by the radian frequency) and force signals
and then combining the bin lines corresponding to each 1/3-octave band.
Each guitar was suspended in a metal framework by the neck and lower bout with
soft elastic strings resting horizontally at 1.20 m from the ¯oor. Several tests were
conducted in which the guitars were interchanged, di€erent attachments for the
exciter tried, impedance head to guitar contact varied, etc., until the best ®nal
arrangement, that which resulted in a higher signal to noise ratio at the input and
higher sound radiation without noise from rebounds of the contact head, was
found.
190 R.R. Boullosa et al./Applied Acoustics 56 (1999) 183±197

The top plate was excited at the bridge with a vibration exciter fed with a band-
®ltered noise signal to restrict the driving frequency within one-third octave bands.
An impedance head was attached to the exciter which had on its end a short (2 cm)
and not too sti€ wire. To this wire a short piece of metal (1 cm) with a small pit the
same width as the string was fastened, which was, in turn, glued in the bridge at the
position of the sixth string. The ®nal measurements were made, in di€erent order, in
the course of two weeks, in each of the four guitars and repeated three times. The
®nal results correspond to the average of these three measurements. The average
temperature during the measurements was 22‹0.5  C and the average humidity was
48‹3%. Fig. 7 shows the results of radiation eciency for each of the four guitars.
Fig. 8 shows the total radiation eciency below and above 1 kHz. Note that to
obtain the total eciency above and below 1 kHz, care must be taken to account for
the fact that the bands used have proportional widths and thus the total eciency
must be weighted by this factor. The total eciency was obtained with

P
N
En fn
Etotal ˆ nˆ1N …2†
P
fn
nˆ1

where En is the eciency in band n and fn is the bandwidth of the corresponding
one-third octave band.

Fig. 7. Radiation eciency in 1/3-octave bands from 100 Hz to 3.15 kHz, for the four guitars.
R.R. Boullosa et al./Applied Acoustics 56 (1999) 183±197 191

Fig. 8. Total radiation eciency in the one-third octave bands (a) from 100 Hz to 800 Hz and (b) from 1
to 3.15 kHz.

5. Subjective evaluation

The guitars were evaluated subjectively in two di€erent ways. One based on `blind'
listening tests by a general audience and another based on direct inspection and
playing by guitar teachers and students.

5.1. Listening tests

The listening tests were carried out in a small lecture room with a capacity of 120
seats; the guitars were played behind a dark curtain, limiting subjects listening to the
guitars to purely acoustical clues as a basis for their evaluation. A short musical
fragment, about 10 s long, was played in succesion on two of the guitars; the subjects
were then given the task of deciding which sound they liked most, (A) that of the
®rst guitar, (B) that of the second, or (X) none. Pairwise comparisons like this were
done on a randomized series of all the 12 possible pairings of the four guitars
(including di€erent orderings of the same pair of guitars) and four control pairs in
which the same guitar was played twice. This gives a total of 16 pairs using the same
musical fragment. Two di€erent musical fragments were used during the tests, these
being the ®rst few measures of a Tarantella in A minor by Fernando Sor and of the
Prelude from Suite BWV~1006 in E major by Johann Sebastian Bach. Using each of
these fragments in the 16 pairs resulted in a total of 32 pairs to be evaluated during
the listening tests. These were carried out on di€erent dates in two separate 40 min-
ute sessions in which the 32 pairs were evaluated; the guitars were played by two
192 R.R. Boullosa et al./Applied Acoustics 56 (1999) 183±197

di€erent players, the same during each session. A total of 51 subjects participated
voluntarily during the two sessions (a few of the subjects participated in both). All
the subjects consider themselves to have normal hearing and their personal involve-
ment in musical activities range from non-musicians to professional guitarists.
The results are shown in Figs. 9±11. Fig. 9 shows how votes are distributed by
option (ABX) during the tests. A reasonable low proportion of X-votes was
obtained which gives con®dence to the relevance of the results. On the other hand,
the proportion of X-votes is not altogether negligible and may re¯ect the diculty of
the test which was expressed by some of the subjects. Also interesting is the fact that
the proportion of B-votes is clearly shown to be larger than the proportion of A-
votes. This seems to suggest that a short-term memory e€ect is taking place, in
which listeners tend to prefer the second guitar (the more recently played) over the
®rst, in spite of the fact that listening to each pair lasts less than about 40 s.
Figs. 10 and 11 show di€erent ways of measuring Subjective Quality calculated as
a percentage of the number of votes obtained by each guitar, and where 100%
means a guitar that obtained the maximum number of votes possible in each case.
The presence of X-votes, and the elimination of votes in Fig. 11(b), implies that the
values of Subjective Quality do not add to 100%. Fig. 10 shows the percentage of
votes obtained by each guitar when presented (A) ®rst or (B) second in a pair. Dif-
ferences between the guitars are less clearly seen when they appear ®rst in a pair,
which is consistent with the previous observation regarding the bias towards the
second guitar in a pair. Fig. 11 shows percentages of the total number of votes and a
special count in which ambiguous votes have been eliminated (a subject emits a pair
of ambiguous votes when he/she prefers di€erent guitars in reversed presentations of

Fig. 9. Subjective quality: (a) votes by option in all pairs; (b) votes by option in control pairs using the
same guitar.
R.R. Boullosa et al./Applied Acoustics 56 (1999) 183±197 193

the same pair of guitars). There is a signi®cant di€erence (about 50%) between the
total voting and the unambiguous voting which is surely due to the bias towards the
second guitar mentioned above. In any case, the unambiguous voting shown in
Fig. 11(b) can be taken as the more reliable result from this test.

Fig. 10. Subjective quality: (a) votes obtained as the ®rst guitar in a pair; (b) votes obtained as the second
guitar in a pair.

Fig. 11. Subjective quality: (a) total number of votes; (b) total number of unambiguous votes.
194 R.R. Boullosa et al./Applied Acoustics 56 (1999) 183±197

5.2. Playing tests

The playing tests were carried out in a small meeting room. Guitar teachers and
students were invited to spend 20 min freely and privately inspecting and playing the
guitars, and to ®ll out a form in which they sorted the guitars according to the fol-
lowing aspects: Appearance, Tuning, Timbre, Volume, Comfort and Quality. A
total of 22 guitarists participated in the tests. Each guitarist produced six lists of the
guitars, sorted from the best to the worst in each aspect. A subjective rating was
calculated by assigning 4 points to ®rst place, 3 points to second place, 2 points to
third place and 1 point to last place. The total number of points obtained by each
guitar in each aspect was calculated and then normalized to give 100\% for a guitar
rated best by all players.
The results are shown in Figs. 12±15. The rating of the guitars shows a great
consistency among the di€erent criteria. Fig. 15 shows that the combined rating
according to the ®rst ®ve aspects coincides very well with the rating of the overall
subjective quality. The rating of the guitars is also consistent with the results of the
listening tests.

6. Conclusions

A summary of the results obtained is shown in Table 1, where the guitars are
sorted by subjective quality, tuning errors and radiation eciency. The table shows
that, although the sortings di€er, at least the best rated guitar is also characterized

Fig. 12. Subjective rating of Appearance; (b) subjective rating of Tuning.


R.R. Boullosa et al./Applied Acoustics 56 (1999) 183±197 195

as having the least mean tuning error and the greatest radiation eciency. This gives
limited evidence (given the range of the present study) that the tuning properties and
the radiation eciency of a classical guitar are somewhat related to the subjective
quality. Stronger evidence for this statement requires, however, further study.

Fig. 13. (a) Subjective rating of Timbre; (b) subjective rating of Volume.

Fig. 14. (a) Subjective rating of Comfort; (b) subjective rating of Quality.
196 R.R. Boullosa et al./Applied Acoustics 56 (1999) 183±197

Fig. 15. (a) Subjective rating of ®ve aspects: Appearance, Tuning, Timbre, Volume and Comfort; (b)
subjective rating of Quality.

Table 1
Guitars sorted by subjective quality (decreasing), tuning errors (increasing) and radiation eciency
(decreasing)

Subjective Mean tuning Max tuning Max interval Radiation


quality error error error eciency

FOB FOB CHF APL FOB


RRB CHF APL CHF APL
CHF APL FOB RRB CHF
APL RRB RRB FOB RRB

Acknowledgements

We wish to thank the following people: Clara Alvarado, Maestro Eloy Cruz,
Francisco GoÂmez and Antonio GuzmaÂn. We also wish to thank members of Centro
de Instrumentos UNAM who took part in the listening tests, the guitar teachers and
students of Escuela Nacional de MuÂsica UNAM, and specially to Maestro Julio
Vigueras for his support during the playing tests.

References

[1] Jaroszewski A, Rakowski A, Sera J. Opening transients and the quality of classical guitars. Archives
of Acoustics 1978;3(2):79±84.
[2] Jansson EV. Fundamentals of the guitar tone. Journal of Guitar Acoustics 1982;6:26±41.
R.R. Boullosa et al./Applied Acoustics 56 (1999) 183±197 197

[3] OrdunÄa-Bustamante F. RelacioÂn entre la calidad subjectiva y algunos paraÂmetros acuÂsticos de la


guitarra claÂsica. Tesis de licenciatura, Universidad Nacional AutoÂnoma de MeÂxico, March 1987.
[4] OrdunÄa-Bustamante F. Experiments on the relation between acoustical properties and the subjective
quality of classical guitars. Journal of the Catgut Acoustical Society (Series II) 1992;2(1):20±3.
[5] Gough CE. The theory of string resonances on musical instruments. Acustica 1981;49:124±41.
[6] Suzuki H. Vibration and sound radiation of a piano soundboard. Journal of the Acoustical Society
of America 1986;80:1573±82.
[7] Lai JCS, Burgess MA. Radiation eciency of acoustic guitars. Journal of the Acoustical Society of
America 1990;88(3):1222±7.

Das könnte Ihnen auch gefallen