Académique Documents
Professionnel Documents
Culture Documents
THE EFFECTS
OF CONCERT HALL ACOUSTICS
ON VOCAL PERFORMANCE
of
by
____________________
Thomas M. Guyette
Approved:
__________________________________
Professor Douglas Weeks, Project Advisor
2
Abstract-
The exploration of the effects of concert hall acoustics on musical performance is one of the most
elusive areas of acoustical research. To expand the body of knowledge in this field,
volunteer professional and amateur singers were placed in a Digital Signal Processing
survey of overall impressions was administered to each singer after every performance.
The recordings were later analyzed by a panel of musicians. Analysis of the surveys
revealed that although musicians are often very aware of their environment, they do not
necessarily know how it affects their performance and mindset. Hypotheses about
description of the procedure and equipment used to create an artificial acoustical environment within an
anechoic chamber, describes the recording and analysis techniques used to formulate conclusions from
the performance data, and offers suggestions for future research in the field.
My thanks goes to Professors Richard Campbell and William Michalson for their patience and
indulgence regarding another concurrent project. Also of note was the sage advice and guidance of
explained in detail how the Lexicon CP1 works, and took time out of an especially busy day to
encourage me.
Special thanks should also be given to Bose Corporation, of Framingham Massachusetts, for the
generous contribution of equipment and space for conducting the experiment. My gratitude goes
especially to Doug Kramer, who manages the anechoic chamber, Charlie Barker, who manages the
equipment, and Richard Saffran, who manages to always be around when he is needed.
This research project would not have been possible without the time and work donated by seven terrific
musicians: Erica Curran, who played the recorded accompaniment for "America The Beautiful"; Ray
Bauwens, who tested the chamber in its initial stages of design and participated in the experiment; Jean-
Pierre Trevisani and Wendy Manas, who performed in the finalized set of environments; Rebecca
O'Brien and Christina Freeman, who performed in the experiment and analyzed the performances of
4
other musicians; and Doctor Douglas Weeks of the WPI Humanities and Arts Department, who advised
Abstract ii
Table of Contents iv
Introduction 1
Literature Review 4
HISTORY OF ACOUSTICS 5
Reverberation Time 5
Frequency Characteristics 7
Sound Distribution 7
Early Reflections 7
Echo 8
Isolation 9
Experimental Procedure 15
GOALS 15
METHOD 15
Subjects 15
Music 17
Equipment 18
Environments 21
Procedure 22
6
(Table of Contents, Continued) Page
Analysis of Data 24
METHOD OF ANALYSIS 24
Environment 1: "Standard" 26
Environment 6: "Anechoic" 50
Environment 9: "Reversed" 64
Performer 1: RB 73
Performer 2: RO 82
Conclusion 95
7
(Table of Contents, Continued) Page
Subjects 100
Equipment 101
Procedure 104
Questionnaires 105
ANALYSIS 107
BUDGET 108
and musical communities on the topic. Traditionally, engineers designed halls, often with unplanned
acoustical properties, and musicians performed in them. There was very little communication between
the two groups. Thankfully, within the past thirty years, the intangible line separating musicians and
acousticians has been fading. Unfortunately, despite this newfound cooperation, the subdiscipline of
hall acoustics which requires the most cooperation between the two groups--the study of the effects of
acoustics on performance--has not advanced as dramatically as the rest of the field. More research is
needed in this area so that the distance between acousticians and musicians can be reduced further.
The first experiments in subjective room acoustics took place in the twentieth century, by W.C. Sabine in
the 1920s, and Leo Beranek, Manfred Schroeder, and others in the 1960s. These experiments were
extremely innovative in their equipment and experimental technique, and broke new ground in trying to
determine what concert halls audiences preferred. From these successful experiments grew a new
Subsequent research in this field was not limited to determining which halls audiences preferred.
Instead, experiments were designed to determine the specific characteristics that made the halls
These experiments were instrumental in understanding the preferences of audiences for certain
measurable acoustical characteristics. Even more importantly, they employed revolutionary techniques
experiments became more reliable; And, as more listener-based experiments were done, more hall
measurement
9
techniques were developed. The process fed upon itself. However, despite the importance of these
studies, they fell short in that they left one important piece of information untouched: the preferences of
the performer.
Subjective impression experiments showed conclusively that audiences have preferences for certain hall
acoustical characteristics. It would be reasonable to assume that musicians must have equally strong--or
stronger--preferences, because the singer, unlike the audience, has an active feedback relationship to the
hall. The musician performs, listens to the result, adjusts, listens again, adjusts, and so on. Audiences
need only listen and enjoy. Thus it would seem logical that the musician, rather than the passive listener,
should be the focus of acoustical preference experimentation. Despite this, upon reviewing current
acoustical literature, one sees that very little is known about the musician's interaction with his
acoustical environment. The literature review for this paper contains a single paper on the topic. And
that paper was recent enough that the data had not fully been processed at the time of publication.
The primary goal of this project, then, is to fill the gaps in the literature on the topic of the effects of
concert hall acoustics on the musician. The focus of the experiment will be singers, for reasons which
are discussed in the body of the paper. The information contained herein should benefit acousticians and
musicians equally. Acousticians designing halls will now be able to consider the musicians who will be
performing in them, and will be able to design with those musicians' needs clearly in mind. Musicians
performing in these halls will be able to put forth their best performances, because they will not have to
In the past, probably the single largest limiting factor in the determination of the effects of hall acoustics
on the performer was a limitation in available technology. Until recently, it was virtually impossible to
10
create a variable acoustical environment. Doing so required large pieces of moving paneling, padding,
and carpeting, which made the procedure expensive and impractical. Thus, the only possible way of
exposing experimental subjects to different environments was to play them "canned" pieces of music,
recorded in different environments. Electronic processing of a live performance in real time was
prohibitively expensive. Bringing an entire audience and performing group into a dozen different halls
is impractical. Until recently, money and practicality played a strong role in the stagnation of the field.
Recent advances in Digital Signal Processing technology make it possible to alter the acoustical
characteristics of a room at the press of a button. Many companies have made electronic devices
specifically for that purpose. Among them are Lexicon Corporation, SigTech, and Stell Acoustics.
Technology has finally caught up to the needs of acousticians, and experimentation in the field has
become possible.
"Possible" does not necessarily translate to "Easy." A single project could never completely characterize
a field which has evaded acousticians for centuries. Therefore, this paper should be viewed as an
exploratory foray--a preliminary piece of work, designed to determine what research needs to be done,
rather than to conclusively prove a hypothesis. Hopefully, this small project will be picked up by the
The following section comprises a review of much of the available literature on concert hall acoustics, to
bring the reader up to date on the field. Although every effort was made to be as thorough as possible,
Just over 130 years ago, Hermann von Helmholtz wrote a landmark work on the physics of acoustics,
On the Sensations of Tone (Campbell and Greated 2). It was at that point that acoustics graduated from
being a concept to being a science. Helmholtz's book opened up discussion and research on all manner
of acoustical phenomena. Acousticians realized early on that the problem of improving concert hall
acoustics was an important one. Unfortunately, that area was doomed to lag behind the rest of the
Concert hall acoustical analysis differs from the other acoustical sciences in that it can not rely as much
on trial and error. Musical instrument design lends itself to speedy progress because prototypes can be
designed, tested, and rebuilt within a matter of days. Concert halls, on the other hand, take a decidedly
longer time to build, and the only available prototypes for new halls are previously built halls (Campbell
and Greated 525). Twenty or more generations of violin can be constructed in less time than a single
concert hall. It is no great wonder that the field of concert hall acoustics lags behind the other areas of
the discipline.
Not only are concert halls more expensive and difficult to build, they are "exceedingly complex" to
analyze, because of the number of parameters involved: analysis must include the shape of the
enclosure, the properties of the materials used, furniture, characteristics of the sound source, position of
the audience, and a near infinite number of other variables (Olson 266). Until very recently, the
resources needed to analyze this large quantity of data simply did not exist (Olson v).
12
Beyond technological shortcomings, another problem existed that slowed the growth of the science of
room acoustics. Acousticians knew that the acoustics of the room in which music was played influenced
a listener's opinion of performance (Campbell and Greated 525). However, early research in acoustics
was carried out "with much simpler sounds and in laboratory conditions far removed from the ambience
of the concert hall (Campbell and Greated 40)." Stereotypically, acousticians worked in their
laboratories, and musicians kept away. This dividing line plagued the science of musical acoustics for
centuries, until the twentieth century. It is fortunate that musicians and acousticians have finally
managed to come closer together, and the science of musical acoustics is growing because of it.
HISTORY OF ACOUSTICS
Before describing the technology and cutting edge research in acoustics, it is necessary to explain the
underlying principles and tenets of general concert hall acoustics. Following is a brief, non-technical
discussion of some characteristics of concert halls which have been of particular interest to acousticians
in the past.
Reverberation Time
In conversations among musicians about acoustics, reverberation is almost always the first attribute
mentioned, probably because it is the most familiar (Gade 18). The presence of reverberation
"considerably alters the quality of ... music (Backus 168)." Some musicians even maintain that
reverberation time at middle frequencies is the "most important" musical factor to be considered
when designing a performing hall (Geerdes 51). Deciding on a reverberation time, or "T60" when
designing a hall is not a simple chore. In the words of John Backus, in The Acoustical Foundations of
Music, "the optimum reverberation time for an auditorium is . . . a compromise between clarity on one
hand and a satisfactory sound intensity on the other. For a given auditorium the best reverberation time
13
will depend on the use for which it is designed . . . (170)." Backus says that a hall with too large of a
T60 is analogous to a piano with a stuck sustain pedal: individual notes become obscured and muddy
(168). A hall with too short of a T60 will sound dry (Gade 15), or "dead (Backus, 170)."
Charts have been published which show ranges of T60 for certain applications, but different charts show
conflicting information. For example, Campbell and Greated list 1.3 to 1.6 seconds as the optimal T60
for Opera (536). Geerdes lists 1.5 to 1.8 seconds (51). Theaters have "optimum T60s" ranging from 1
second to 1.4 seconds (Geerdes 51; Campbell and Greated 536). Despite over one hundred years of
acoustical research, musicians and acousticians still have not unanimously agreed on such a deceptively
simple characteristic.
To add even more complication to the issue of optimum reverberation time, Campbell and Greated point
out that that T60 can be different for different pitches, and that this distribution of T60 over the full
auditory frequency range is important. For example, it is generally accepted that 19th-century music
should be performed in a hall in which T60 is smaller at higher frequencies. Baroque music, on the
other hand, sounds better in a hall with a "flat" [constant across all frequencies] T60 (Campbell and
Greated 537). In fact, the "warmth" of a hall is directly related to the ration of T60 at low frequencies
Although the studies on optimum reverberation time do not agree in full, this is not for lack of trying;
RT is merely a matter of personal preference, as are all other hall acoustical parameter preferences.
Because RT has been explored in such depth already, the experimental procedure of this project will not
The mention of the frequency dependency of RT above raises a larger issue of frequency characteristics
of a hall in general. Some halls are "bright," others are "dull." There is a host of other terms used to
describe the frequency content of the hall's response to a performer's instrument. Not much research
was found regarding the effects of frequency-dependent level of return from an acoustical environment.
Therefore this experiment dives quite deeply into the effects of frequency characteristics of a hall on
performance.
Sound Distribution
One room parameter that is very closely related to reverberation time, about which unanimous
agreement has been reached, is sound distribution. Second only to reverberation time, sound
distribution is the most talked about, most widely known characteristic of a hall (Backus 177).
Refreshingly, there is no dissent as to what type of distribution is desirable: it is generally agreed that
sound should be evenly distributed throughout the hall. Unfortunately, that is virtually impossible to
achieve (Gade 21). In fact, there is often a larger difference between different seats in the same hall than
between similarly located seats in different halls (Gade 7). So although there is no disagreement that
even distribution is desirable, so far it seems that this agreement is sadly moot, and acousticians must
This project pays much attention to sound distribution, in the form of the directionality of the return of
the performer's voice, in the form of reverberation and early reflections, from the concert hall.
Early Reflections
Although reverberation time and even distribution of sound were the first parameters of acoustical
spaces to be studied, the development of electronic recording, digital signal processing and directional
15
microphone arrays has made it possible to focus less on the very broad, statistical topics of T60 and
distribution, and more on the specific behavior of a hall in the short interval of time after a sound is
produced, in the first tenth of a second or so--that is, to focus on the relationship between the original,
"direct" sound, and the first reflections from nearby surfaces (Jordan 114).
When a performer on a stage produces a sound, the sound travels, at the same time, to nearby audience
members and to nearby surfaces. It then rebounds from the surfaces, and continues to rebound until it
reaches a listener, or decays to an inaudible level. Thus, a short time after they experience the "direct"
sound from the performer, audience members experience the "reflected" sound from nearby surfaces.
The time interval between the direct and reflected sounds has a very important effect on how the
audience will feel about the sound--specifically, how "intimate" the space is. Recent research suggests
that this time interval should be less than 20 milliseconds (Backus 179; Campbell and Greated 540). It
has also been known for some time that the direction of arrival of the first reflection also affects an
audience's level of pleasure. It is desirable to have the first reflection arrive from the sides, rather than
from above or behind (Campbell and Greated 540; Backus 181; Jordan 115), or the hall will sound "dry
(Jordan, 115)." In this project, some time was devoted to the effects produced by different early
reflection patterns.
Echo
One effect of improperly timed and improperly angled reflections is echo. If one particular reflection is
time-wise and intensity-wise separate from all other reflections, i.e., if it can be singled out by the
human ear, it is called an "echo." Echoes in performance conditions can be disastrous (Backus 179).
Sometimes echoes do not decay after they are heard. Instead, an echo can persist, bouncing back and
forth between two parallel surfaces many times before dying out. Such an echo is called "flutter," and is
even less
16
desirable than single echoes (Campbell and Greated 542). Obviously, then, it is important to pay careful
attention to direction, intensity, and timing of all reflections. Again, this task involves a great deal of
variables, and is therefore overwhelmingly complex. Just as with reverberation and distribution,
acousticians have been forced to make generalizations and compromises. However, there is no counter
to the fact that echo is bad for a performer, and this survey pays no time studying its effects on
performance.
Isolation
Isolation from noise is the one acoustical parameter that is not in debate, and possible to achieve. The
ideal hall would have absolutely no background noise. However, the behavior of the human ear is such
that sufficiently soft sounds absolutely can not be heard (Beranek, 394). Further, it is possible for louder
sounds to completely "mask" softer sounds (Beranek, 404). This masking effect means that there is a
level of background noise that can be considered "low enough" for performance conditions. The only
challenge, then, is to achieve this level: to soften the noise of ventilation systems, to acoustically insulate
This study does not devote any time to isolation. Although lack of isolation is an irritation to any
performer, the environment used in this experiment was not flexible enough to provide any degree of
Obviously there are still mysteries in the realm of room acoustics. In the words of John Backus, ". . . the
subject of auditorium acoustics is by no means completely understood. Research and argument are still
going on, and will continue to do so (181)." Thankfully, the field if acoustics is no longer the exclusive
17
realm of physicists in laboratories: it has become a diverse, interdisciplinary field, comprising not only
physicists but physiologists, psychologists (Campbell and Greated 2), and musicians. Because of this
unification of artists and scientists, some of the most recent advances in the last decade (Jordan 116)
have been in subjective analysis (Olson v)--that is, determination of which measurable attributes of a
concert hall correspond to which subjective impressions. In fact, that research is used as a point of
In "Acoustical Criteria and Acoustical Qualities of Concert Halls," Vilhelm Lassen Jordan asserts, "the
question of correlation between objective measures and subjective impressions of acoustical quality [has
become] more and more urgent." Up to a little over a decade ago, research in subjective perception of
hall acoustics was limited to a few pioneers. W.C. Sabine conducted experiments in 1923 which tried to
determine optimal reverberation times for concert halls. Leo Beranek is credited with doing the "most
extensive" survey of this kind, in 1962, studying more than 50 concert halls worldwide. Both men used
musicians' subjective impressions of halls, combined with objective measurements of these halls, to
determine what, in a hall, was "good," and what was "bad (Gade, 8)."
A "large number of studies" have been done to relate subjective and objective characteristics (Bolzinger,
Warusfel, and Kahle) of halls. Many of these studies, including Sabine's and Beranek's, included
musicians as experts. In fact, musicians can be considered experts in hall acoustics, as they relate to
performance. A recent survey found that musicians quite often communicate with each other about the
shortcomings and advantages of different halls (Gade 40). 81% of musicians surveyed agreed that
acoustics were important to the performer (Gade, 36), and 63% of the surveyed musicians were aware of
gaining the ability to be aware of, and adjust to, the acoustics of a hall (Gade 35). However, less than
30% of the musicians had ever spoken to an acoustician or someone technically familiar with acoustics
18
(Gade 41)! Obviously more needs to be done to open the lines of communication between acousticians
One paper, done by Vilhelm Lassen Jordan in Denmark, established very clearly which numerical
measurements corresponded to certain impressions named by musicians. Jordan defined a specific set of
acoustical parameters which can supposedly uniquely characterize a hall. For example, he defines
"Early Decay Time," or "EDT" as the amount of time it takes an initial sound to die out by 10 decibels
(see Appendix 1), and related it to the subjective characteristic of reverberance. Another example was
the idea of "Deutlichkeit," or Clarity, which corresponded to the amount of energy reaching a particular
spot during the interval from 0 to 50 milliseconds, to the amount of energy reaching that spot after 50
milliseconds. Although this paper was very thorough in making measurable parameters of previously
nebulous ideas, it did not specify what values corresponded to "good" acoustics. That job was left for
later study.
One recent survey, done in 1981 by Anders Christian Gade at the Technical University of Denmark, did
just this. Its goal was to determine what objectively-measurable room parameters corresponded to
which subjective impressions. To this end, an interviewer with a list of questions took a group of
musicians, one at a time, into a room, and asked a battery of survey questions. The survey was aimed to
determine what characteristics musicians found desirable in a hall. In order to analyze the data, the
interviewer had to "translate" the words of the musician into a group of standard terms. It was
determined that all musical impressions could be mapped to seven terms: Reverberation, Hearing-each-
other, Support, Timbre, Dynamics, Time-delay, and Change of Pitch. Reams of statistical analysis
determined what values for each of these characteristics musicians found most important--or at least,
which of these characteristics musicians liked to talk about most. The surveys also supposedly hinted at
what
19
characteristics of playing (tempo, attack, volume, and the like) were affected by which hall
characteristics (reverberation, support, etc). This particular survey had the opposite problem as that
done by Jordan--it gave no values or ranges of values for the parameters mentioned, because, in the
words of the writer, "at the moment so little is known about this field that an interview survey collecting
opinions unbiased by the presence of technical equipment would be of value (11)." This leaves a perfect
niche for the project described in this paper, which combines the survey technique with technical
equipment.
Even research that has been done with musicians, like Gade's, has flaws. Most experiments in this field
deal with musicians as listeners, not as performers, "and may hence fail for the description of the
sensations of the musicians on stage (Bolzinger, Warusfel, Kahle)." Simply put, researchers have not
been using musicians to their full potential: as experts in determining which acoustical characteristics
will encourage a virtuoso performance. After all, "acoustics may be regarded as one of the main
qualities of the musician's working environment (Gade, 7)." One thrust of modern acoustical research,
then, is to determine how performance hall acoustics affect the performer on stage (Bolzinger, Warusfel,
One may wonder: if we have already determined what conditions are desirable for a listener, why would
the musician's perspective be any different? There are several reasons. First, acoustic conditions are
seldom the same on stage as in the audience. Second, musicians' needs are very likely not the same as
the needs of an audience (Gade, 7), because a musician's relationship with a hall is active--it involves a
broadcasting again (Gade, 8). Of all of the studies and research done in the area of musical acoustics,
environmental acoustics on the performance of the concert pianist. Seven professional soloists were
brought into a darkened concert hall with adjustable acoustical properties, and were asked to play a
session that consisted of a few minutes of practice, some short imposed exercises, and some music of
their choice, which had to be all written--it could not include any improvisation. They were asked to
play this sequence nine times. Each time, certain acoustical parameters of the hall were changed. The
hall in which they played was physically alterable, meaning that in order for changes to occur, barriers
had to be moved, the ceiling was raised or lowered, walls' material was changed, and so on. Only the
acoustical properties of the hall were changed. The pianist played on the same piano, in the same hall,
on the same day. The changed parameters were T60, reverberation level, the ratio of direct to reflected
sound at the pianist's ears, and the frequency characteristic of the reverberation. Audio was recorded on
a MacIntosh computer, and the nuances of the pianist's performance was recorded via a MIDI grand
piano. After each of their sessions was recorded, the musicians were asked to answer a brief survey
Analysis of the data is still continuing, but preliminary analysis found that the reverberance of a concert
hall definitely affects the intensity with which a pianist plays. However, because the hall had physical
limitations on alteration of parameters, it is not clear whether it was the reverberation time or the
The correlations of the musicians' impressions were higher than the correlations of audience members,
confirming the suspicion that musicians are more 'in tune' with their acoustical environment. This study
effectively contradicted earlier research, and held that reverberation time has little effect on the
musician's choice of tempo. Even more interesting, one setting was almost unanimously preferred by
the musicians
21
involved, regardless of what style of music they played! These contradictions emphasize the need for
more investigation into the phenomenon of performer/hall feedback. It is possible that early survey-
based research in this area is flawed, either because musicians have difficulty putting their impressions
into words, or they are not always consciously aware of a hall's effects on their performance (Gade, 11).
We have seen that there is no shortage of research dealing with impressions of a listener in a given
acoustical environment. It is also obvious that the research done in this area up to now has flaws, and
would benefit from further exploration. It is already known that room acoustics have a psychological
and physical effect on the performer. Musicians in good environments feel more inspired, can
concentrate better, and do not have to adjust their playing style to compensate for poor acoustics (Gade,
29). As we have seen, though, there is a void surrounding the idea of the specific effects of an acoustical
environment on a performer. Of those papers that did delve into this field, only a "small fraction" are
based on empirical data (Gade, 8). Some of the survey-based studies had problematic question formats:
one only encouraged musicians to pick one acoustical effect as "the most important," and did not bother
to rank the rest (Gade, 23). The one study that did base its results on empirical findings had trouble with
its experimental process, and could not completely isolate specific causes (Bolzinger, Warusfel, Kahle).
Therefore, it is logical to conclude that future research should concentrate in two areas: first, it should
concentrate on determining the effects of concert hall acoustics on a musician's performance, and
The primary objective of this experiment was to determine how certain acoustical conditions affect a
musician's performance. No judgment was made, before the experiment was complete, which
environments were considered "good" and which were considered "bad." Although some results were
expected, the ultimate results were determined empirically from the performances generated under the
The first project objective was to determine what physical and psychological effects an alteration in
acoustical environment can have on a musician's performance. In other words, how the performer
adjusts and how performance suffers under certain conditions, and excels in others, and how musicians
The second objective was to determine whether a musician's impressions of his performance are
accurate. That is, whether a musician would rate his performance the same as an objective audience
member, and how performance hall acoustics affect the musician's ability to judge his performance
objectively.
METHOD
Subjects
Before proceeding, subjects had to be found for participation in this experiment. Because of budget
restrictions, it was necessary to solicit volunteers. It was already determined that the experiment would
rely upon putting musicians in different acoustical environments. The challenge was to determine what
expressive of all instruments (3)." Although this may initially seem a pompous attitude propagated by
singers, there is truth to the statement. The singer is indeed more personally involved with his
instrument than any other musician. If the instrument is hurting, the singer knows it instantly, and
suffers with it. Voices are unique to the individual, and are used (by some better than others) for an
entire lifetime. Also, solo singers are at greater risk than other instrumentalists if they are exposed to a
hostile acoustical environment, because, in Gade's words, "the tendency is to 'force' sound, which
deteriorates the timbre of the sound (32), and can even cause damage to the voice."
Not only are singers more "in tune" with their voices, they are bound to be more personally involved
with the hall in which they are singing because voices, unlike pianos, have no sounding board. The hall
is their sounding board, an extension of their instrument, and therefore affects them more profoundly
than any other instrument (Gade 37). Quoting Carl Shalk, Lovelace writes "the building itself is an
instrument."
Even beyond these very effective points, practical experimental concerns made vocalists a logical
choice: the human voice is portable. Unlike a piano, it can easily fit through doorways. It does not
require electrical power. Its only disadvantage is that it may tend to wear out more quickly than other
instruments--a topic which is described in the data analysis, later in this paper.
The combination of these reasons led logically to the conclusion that the participating musicians for this
experiment would all be singers. That is a rather broad description, however. Opera singers certainly do
not fit in the same category as rock singers. Sopranos certainly do not fit into the same category as
bass/baritones. It was decided that variety would be a strength in determining the demographics of the
participants, and thus the singers invited to participate were gleaned from a variety of styles, ranges,
24
timbres and experience levels. Because previous research has shown that sensitivity to room parameters
is a learned trait in musicians (Gade 28, 35), and that solo singers are more aware of what constitutes a
good performance, (Gade 2, 29) all of the singers chosen had at least some solo performance experience.
The initial contact letter sent to the performers has been duplicated in Appendix 2.
The participating singers in this experiment are Ray Bauwens (referred to as RB), a part-time
professional tenor from the Worcester area with extensive classical and opera solo experience; Christina
experience with the choral environment, a bit of musical theater, and a bit of solo work; Rebecca
O'Brien (RO), a part-time professional soprano hailing from the Boston area with many years of
classical and operatic solo experience; Jean-Pierre Trevisani (JP), a professional tenor from WPI who
has participated in the Metropolitan Opera's Young Artists program; and Wendy Manas, a soprano from
WPI who has years of choral experience, some musical theater, and some popular/patriotic solo
experience.
Music
One challenge in designing the experimental procedure was the decision of what type of music would be
sung. The songs had to be familiar, so that memorization of the pieces would be unnecessary, since
25
sheet music in the chamber would have provided extra acoustical reflections to the singers and skewed
the data. The music obviously had to be vocal, have some sort of artistic merit (i.e., Happy Birthday To
You wasn't a viable choice), be singable by all vocal parts, and with the exception of the a cappella piece,
had to have a simple piano accompaniment. Although it would have been desirable to have music which
was
26
phonetically balanced--that is, music that has a statistically even distribution of English phonemes--no
Gade chose classical pieces for his work, because of the number of performers involved, and because of
the dependence of classical music on room involvement (11). Rock music, for example, would not be
appropriate, because rock bands often make use of electronic equipment, and therefore are not as
sensitive to room acoustics. This study thrust more toward patriotic music: America the Beautiful was
sung with accompaniment, and The Star Spangled Banner was sung a cappella. Each singer also
Ray Bauwens's chosen piece, as well as Jean-Pierre Trevisani's, was Giordano's Amor Ti Vieta.
Christina Freeman/White sang Rutter's Pie Jesu. Rebecca O'Brien performed Barber's Sure On This
Shining Night, and Wendy Manas used Rock-A-Bye Your Baby, arranged as part of a medley of
twentieth-century American songs. All of the soloists provided their own piano accompaniment on
cassette tape.
Equipment
Unfortunately, it was not possible to transport the musicians between real performance halls. The next
best option, then, was to design an artificial acoustical environment which could be changed with the
touch of a button. Ideally, this room would have been alterable by addressing the parameters discussed
in the Literature Review section of this paper: reverberation time at different frequencies; reverberation
intensity at different frequencies; distribution; early decay time; early reflection directionality,
concentration, and intensity; isolation, and so on. Unfortunately, limitations in budget prevented the
creation of a very high-tech environment in which to work. An in-depth description of the chamber that
was used, which evolved from the equipment which was available, follows.
27
The first requirement in designing an artificial environment is to have a place in which to build it.
Obviously, the room in which it is implemented must have as little "acoustical personality" as possible:
response to all sound within fifty miles of the WPI campus: the
used for research by Bose engineers, in exchange for a copy of The Anechoic Chamber at
Bose Corporation
this document.
With a place to build the environment, the next step was to put together the electronics for acquisition,
processing, and rebroadcast of the singer's voice in the chamber. Several Digital Processing (DSP)
systems were considered, among them the Lares system from Lexicon, the CP1 from Lexicon, the
Ensoniq DP4, and the Yamaha DSP1. Because Bose Corporation already had a Lexicon CP1 effects
processor on site, the CP1 became the basis for the artificial environment.
To acquire the singer's voice, the system used two Shure SM57 microphones (practically the industry
preamplifier, all borrowed from the WPI Instructional Media Center. The stereo mixer output fed the
CP1 inputs directly. The singers were also fitted with a pair of glasses with binaural microphones
mounted over the ears, borrowed from Professor Richard Campbell of Bang Campbell Associates. It
was originally intended that the signal from these microphones would be used as an accurate measure of
the singers' volume in different environments, because, unlike the volume of the signal reaching the
(Main)
151 151
SM57s
151
151
(Left) (Right)
All microphone signals, comprising the two Shure SM57s and the two channels of the binaural glasses,
were recorded on a Tascam DA88 digital audio recording system, also provided by Bose corporation.
The six CP1 outputs drove three stereo power amplifiers, a Crown D-75 for the CP1 "Main" outputs, a
Crown MT2400 for the CP1 "Rear" channels, and a Crown MT600 for the CP1 "Side" channels, all
borrowed from Bose Corporation. All of the amplifiers fed Bose 151 loudspeakers, comprising a matrix
of six speakers: two directly in front of the singer, two to the sides, an two at the rear (see Fig. 1).
30
The singer's accompaniment was provided by a Bose Roommate speaker, placed on a crate in front of,
and to the left, of the singer. The Roommate was fed by a Califone tape deck, which was operated
outside of the chamber by the person conducting the experiment. Exact adjustment of the volume of the
accompaniment was impossible, because of the variations between tapes. Every effort was made to keep
the accompaniment volume as constant as possible. If accompaniment volume was a factor in a change
of performance style, that fact will show up in the analysis of the a cappella segments.
A comprehensive list of all of the equipment used in this project, including cables and tape, can be found
in Appendix 3.
Environments
With an adjustable environment set up, the next step was to determine the specific settings in which the
singers would perform. A great deal of thought was spent on deciding how many environments were
necessary to gather a reasonable amount of data, without tiring the singers. It was finally decided that
ten environments provided enough variation in environmental parameters to make a good "first cut" at
this sort of research. A singer's session would consist of all ten environments, plus three repetitions
Ideally, the parameters to be varied would have been expressed in terms of standard acoustical
measurements--i.e., "-3dB in the 1kHz octave band, -10dB in the 2kHz octave band, etc.," or "RT of 3.2
seconds at 500Hz, 2.5 seconds at 1kHz," and so on. However, equipment limitations made it necessary
to express room parameter changes in terms of equipment settings, such as volume levels, power
amplifier on/off, LIVENESS, etc. A full list of the characteristics of each setting, and a list of the order
in which the settings were presented to each performer, can be found in Appendix 4. Setting 1 was the
31
"Standard" setting, from which all of the other settings were derived by making one or two
modifications to the equipment setup. Setting 2, the "High Enhanced" setting, was created by turning
the Mackie High EQ to maximum, and adjusting the CP1's ROLLOFF parameter to 14.9kHz, to create a
large high frequency response. Setting 3, "Most Live," was derived by changing the CP1's LIVENESS
setting to 6, the highest setting. Setting 4, "High Inhibited," was produced by turning the Mackie
Mixer's High EQ to its lowest point, and changing the CP1's ROLLOFF parameter to 329Hz, resulting
in a muted high frequency response. Setting 5, "Less Presence," was set up by turning the EFF LEVEL
on the CP1 to -20dB from the standard -14dB, effectively reducing the level of the room's return by half.
Setting 6, "Anechoic," was the dead anechoic chamber, with all equipment turned off. Setting 7, "Fan
Room" was created by changing the ROOM SHAPE parameter on the CP1 to "FAN," which created a
different early reflection pattern. Rooms 8 through 10, "No Sides," "Reversed," and "Sides Only," were
created by turning the power amplifiers on in different configurations, simulating different directional
patterns.
It should be noted that none of the room settings sounded perfectly natural. Due to equipment and time
limitations, it was necessary to settle for approximations of real environments. The differences between
the artificial environment settings was an exaggeration of the differences which exist between real
concert halls. The results of this study should be read with this in mind. Even if the results are not
predictive of exactly what occurs in natural environments, they at least add to the body of knowledge
Procedure
With singers gathered, an environment constructed, and a plan of attack laid out, the next task was to
gather data. For all singers, an introductory script (which can be found in Appendix 5) was read by the
32
author, the singer was sealed into the chamber, and the first performance began. The singer was
provided
33
with a summary of the script's instructions to follow once in the
chamber was opened up, the singer exited, drank the amount of
water allotted them, filled out the three survey sheets (Appendix
7), and repeated the process. When the singer became tired or
Ray Bauwens was the original test subject for the procedure, and because of limitations in the number of
singers available, he was called back to do a "real" experimental run, complete with recording. In the
initial testing of the procedure, he did not encounter any discomfort with performing three songs thirteen
times. However, without exception, the five singers all described fatigue setting in after six or seven
trials. In fact, Jean-Pierre Trevisani refused to continue to sing after trial number six, so the data from
his trial is incomplete, and has skewed the overall data in favor of some of the environments in which he
did not sing. Regrettably, this kind of experience can be expected when dealing exclusively with
volunteers.
Once all of the data had been gathered, a panel of three musicians was assembled to analyze the
recordings of the singers' performances, to determine if their impressions of their performances were
valid. The forms used in the evaluation of the performances can be found in Appendix 8.
The following two sections describe the analysis of the data gathered in this experiment. The first half,
Environment by Environment, is a compilation of the five singers' impressions of each environment, and
contains some interesting reactions on the part of the singers. The second half, Performance by
Performance, is a summary of the judgments of the musician panel on each singer's performance.
34
Analysis of Data-
METHOD OF ANALYSIS
Part One is an environment-by-environment analysis of how the performers felt in the different
acoustical environments. Because there were not enough performers to achieve statistical significance, a
meta-statistical approach was used. In the case of open-ended questions, such as "Acoustically, what
would you say this setting lacked," two types of answers made it into the final report: those that were
agreed upon by two or more performers, and those that were contradicted. In the former case, the
agreeing performers are quoted together (two trials of the same singer in the environment are counted as
agreements) under a "+" followed by a summary of the singers' point of agreement. In the latter case,
the disagreeing performers are listed separately from the agreeing ones, under a "-" followed by a
summary of the singers' point of contradiction. Answers which did not fit into one of these two
categories were discarded, except in rare cases. The answers and summaries are followed by an author's
analysis of the phenomenon being encountered. The letter/number codes of the form XX# denote the
singer's initials and the trial number for that particular singer.
All judgments as to what performers were describing, and whether their statements agreed or disagreed
with each other, had to be judged as objectively as possible by the author; however, because musical
terminology is sometimes holistic, creative judgment was sometimes necessary to interpret what the
Answers to numerical questions ("Rate this environment," and the like) were statistically analyzed in
terms of arithmetic mean, median, and high and low extrema. Further analysis is unnecessary, because
of the statistical insignificance resulting from the limited number of data points.
35
Part One of the analysis aims to get a sense of the singers' impressions of the environments in which
they performed. This can be used as a point of departure for a second pass over the data to determine
how well they performed in the environments. Part Two does just that, using the impressions of
Because the questions asked of the listeners were different from the questions asked of the singers, the
data could not be judged like the data of Part One. Instead, the focus is on general trends in correlations
between comfortability level and certain types of performance problems, such as glottal stops, pitch, and
vocal timbre. Because of the large number of questions involved, trends are extraordinarily difficult to
find in the data. However, the graphical method used to present the data in Section Two makes it
possible to find trends visually, rather than relying on on-paper numerical analysis. Further, the author
offers to search his data for further trends, if the reader is curious about other types of analysis.
36
ANALYSIS, PART ONE: ENVIRONMENT BY ENVIRONMENT
All environments were produced using the Lexicon CP1 effects processor. Because the only room
measurement technique available was MLSSA, and because MLSSA failed to return valid data about the
characteristics of the environment (possibly because of time variant algorithms used inside the CP1
processor), the environments are subjectively labeled ("High Enhanced," "Most Live," and so on), and
technically designated by the settings of the equipment used to produce the artificial environment.
Using these descriptions, a full reconstruction of the performance conditions of each environment is
possible.
Environment 1: "Standard"
ROLLOFF 2.9KHz
PAN 28
POSITION 127
ANGLE 57
This setting was chosen as a midway point between extremely live environments and extremely dead
environments. The other nine environments are exactly the same as this one, with a single parameter
change either in frequency, directionality, or volume level of the environment's acoustical return.
37
Acoustically, what would you say this setting lacked? How did you notice it?
RB1: "I would lose the sense of sound returning to me in passages in the upper part of my range"
RB5: "It lacked the brilliance of the previous setting -- I didn't sense a brightness of sound"
There seems to have been too much high frequency content in the room's return. Throughout all of the
trials, many of the singers' comments indicated that the artificial acoustic environment tended to sound
unnatural. It is also likely that at times the equipment was "saturating." That is, it reached a point where
it was trying to produce more high-frequency signal than it could. This phenomenon could be the result
of using the CP1 effects processor, which was originally designed for use in home theater systems. The
result is similar to the high frequency feedback found in poorly designed Public Address systems. This
observation has no relevance to real concert halls that do not use amplification.
Acoustically, what would you say this setting did not lack? How did you notice it?
+ Resonance
the reason, with her comment, "the middle range felt very good." The singers were probably noticing
that, although the high frequency component of their voices was saturating the equipment, the "power"
part of their voices--the fundamental frequency of their voices' harmonic series--was returning to them,
How do you think you adjusted, in order to sound your best in this environment?
+ Held back
RB1: "When I noticed a decrease of reverb or resonance in the upper range, I tried to back off
the sound"
RO6: "I brightened the sound -- I felt that improved the pitch."
In all cases under "+," the singers are describing holding back on the high frequency content of their
voices. RO mentions in another question that that she "darkened" the sound. It would seem that some
of the singers tried to correct for the environment's shortcomings by changing their vocal timbre or
The contradictions seem to indicate despair; that is, the singers have realized that the environment is not
an ideal one, so they have decided to "cut loose" and roll with whatever the environment can provide for
39
feedback. The singers seemingly realized the environment was not high-frequency friendly, and tried to
If you could change one specific acoustical property of this environment to make it better, what would it
+ Less Reverb
RO6: "I would take away some of the resonance -- it makes me use too much voice -- I would
also take some of the reverb -- if you hear lower pitches sometimes you can get into a
snowball effect."
CF6: "Less REVERB. Gad. I don't want to know every time I messed up."
- More Reverb
RB5: "I would give it a bit more reverb - I would like the feeling of the sound lingering a bit
longer."
Interestingly, the only singer who wanted more reverberation was RB, the only male who sang in this
environment. I strongly believe this is due to the frequency envelope of the reverberation; that is, there
was quite a bit of reverberation at higher frequencies [hence the saturation], and not enough at lower
frequencies. This reinforces the statement which was made in the introduction about ideal RT being a
The room was very bright, like a rooms with smooth stone walls tend to be.
How do you think a member of the audience would rate your performance in this setting?
+ Full, balanced
- Not balanced
This is a terrific case study in the ability of singers to adjust to performing in almost any environment.
RB had sung in four different artificial settings before he made the contradictory comment that Setting 1
was well balanced. Most likely, compared to the setting he had just come out of (the High Enhanced
setting), the high end saturation was much less offensive to his ears than it was upon first entering the
chamber. Thus his opinion was very much dependent on the context in which he heard the setting.
How would you critique your performance in this setting? What mistakes did you catch yourself
in this environment. In no cases did they hint that they were very pleased with their performance in
these conditions, possibly because, as CF continues to point out, they heard too much: too much of their
own mistakes, too much of their 's' sounds, and possibly even too much of their own breathing. This
brings up a point: perhaps being in circumstances where one hears too much of certain things (i.e.,
which has 'spikes' in certain frequency ranges) can be extremely dangerous to a performance.
The large range of opinions expressed in the numerical data lead me to wonder whether numerical
survey techniques for situations like these is possible, valid, and consistent. It seems as though the
reason this
42
field has been so elusive in the past is that numerical techniques are so difficult to tie into subjective
impressions.
However, for the sake of getting an overall view of all of the environments, all straight numerical data
will be included in the survey process. Perhaps by the end of the analysis, a trend will emerge that will
Standard settings, plus Mackie mixer High EQ all the way up, CP1 ROLLOFF 14.9KHz.
This environment was designed to be very responsive at high frequencies, while keeping the same
amount of responsiveness at middle and low frequencies. The result is a very sibilant environment.
Acoustically, what would you say this setting lacked? How did you notice it?
+ Reverberation
Generally, the artists were very conscious of any imbalance over the full frequency range. Evenness
seemed to be something that they wanted, more than any other characteristic. This is a recurrent theme.
43
Regarding RB's comment that it did not lack reverberation or resonance--he may be referring to the large
amount of high frequency feedback in the room, whereas the may be commenting on the lack of low
frequencies. In all cases, the artists are commenting upon the same phenomenon, in different terms.
Acoustically, what would you say this setting did not lack? How did you notice it?
B4: "I noticed the consonants being reflected back (extremely rare)"
+ Vocal feedback
F7: "Resonance."
All of the singers, off the record, commented on the artificiality of this environment. The author agrees
wholeheartedly that this was the least natural of all of the settings, and is probably not a good example
from which to judge of the effects of concert hall acoustics on a singer. It did evoke some interesting
opinions from the singers, and its results are included in this paper for that reason.
First, it should be noted that feeding singers "technical" terms like resonance in the survey was a
mistake, because they applied the terms to unrelated effects. In this case, CF and WM are commenting
that they are receiving a lot of their voice in a non-delayed reflection from directly in front of them,
rather like they are singing in front of a marble wall, and they mistake this for resonance, which is
altogether different.
This confusion of technical terms can be accounted for in two different ways: future surveys should
either
44
provide definitions of technical terms in the introduction so that the singer is not tempted to provide
their
own definition of the term if it is not familiar or should forego the use of technical terms altogether.
How do you think you adjusted, in order to sound your best in this environment?
O9: "Darkened the quality - i.e., more room further back in my throat"
M7: "couldn't be as free with my singing - had to concentrate on how it sounded in the room."
+ Belt
There are several different phenomena at work here. RO has paid attention to the frequency imbalance
of the setting, and has adjusted her voice to fill out the muted lower frequency range. WM has become
preoccupied with the sound of her voice being reflected directly back at her. RB and CF have opted to
It is possible, at high volumes, to get an adrenaline rush from the sheer intensity of the room's return on
a vocal investment. This "fight or flight" response is similar to the familiar adrenaline reaction to a
fireworks display, loud rock concert, or roller coaster. It is possible that RB and CF have tapped into
Although this sounds like a rather bizarre way of explaining their survey answers, the author has
experienced this phenomenon, and is convinced that it would be a particularly interesting focal point for
If you could change one specific acoustical property of this environment to make it better, what would it
+ Lower the high frequency reverb, give some lower frequency support
RO9: "I would add some resonance and tone down the reverb."
WM7: "Put in more reverb - I like the way my voice sounds with reverb."
- Don't change
Again, the singers are stressing the importance of evenness across the frequency spectrum. CF seemed
How do you think a member of an audience would rate your performance in this setting?
+ Good
RB has used a buzzword, "exciting," that furthers the idea presented earlier -- that of getting "driven" by
Interestingly, the one question that seemed to have the most unanimous agreement was "How powerful
do you feel in this environment." This can be viewed as further support for the 'fight or flight' theory.
Again, the wide range of results lead me to question the validity of numerical surveys for this type of
study. The open form questions seem to be providing a lot more insight into the effects of the acoustics
on the singers than the numerical data. Fortunately, this is to be expected in a first-time experiment in a
Standard settings, plus CP1 LIVENESS 6. The specific effects of the LIVENESS parameter are
unknown. However, the net result was an emphasis of high frequencies in the return signal from the
CP1.
RB2: "This setting lacked a full sound - it was somewhat empty [although with some feeling of
RB8: "Lacked fullness in the bottom range - low notes didn't blossom"
WM2: "The ability to hear my words clearly. Couldn't hear the dynamic style change in my
voice."
This environment, like several of the others, lacked consistency across the full frequency spectrum. This
was partly because of the setting, which was by definition live, and partly because of the setup of the
room, which was prone to a bit of high frequency feedback. CF seemed to enjoy everything that was
Acoustically, what would you say this setting did not lack? How did you notice it?
RB2: "The setting didn't lack fullness in the lower range - speaking voice and low notes elicited a
48
fuller response."
WM8: "Resonance"
Oddly, in some of their trials the singers felt that the environment lacked lower support, and in other
trials they felt that the environment was amply supportive. RB comments that the environment lacked
high support, but had plenty of low support. He later states that it did not lack resonance or brilliance in
the top range. Once again, the singers' impressions are time-variant. Context seems to be having a
profound effect on impressions. When faced with environments that at first seemed unfriendly, they
performed with comfort, confidence, and ease. This suggests a very interesting starting point for further
research: namely, context effects. That is, a study on the effect that a previous acoustical environment
Not only do these results suggest the existence of context effects, it could be inferred that practicing in
an adjustable environment could be considered part of a training regimen for touring singers. RO
commented several times that she would be much more comfortable singing in foreign environments
after participating in this experiment, because she had practiced adjusting (or not adjusting) to what she
was hearing. She further commented that she often felt, while singing in acoustically poor rehearsal
spaces, that her voice was at fault. In the course of this experiment, by singing in ten distinct
environments over the course of three hours, she learned that her voice was in fact consistent, and that
the environments could have a definitive effect on her perceptions of her performance. Perhaps every
singer should have an adjustable environment in which to rehearse, for the purpose of getting used to
RB8: " I tried to fill out the bottom a bit more to compensate for the brightness at the top"
Again, the responses to the questions in this setting seem contradictory, even on the same survey sheets.
It is difficult to interpret what the singers are trying to verbalize. It is possible that at low volumes (i.e.,
speaking), the room's response in the lower range was acceptable, but at higher volumes the
If you could change one specific acoustical property of this environment to make it better, what would it
Although the comments varied, the general feeling was that the setting was not balanced. One
explanation could be that the Reverberation Times at different frequencies was wildly different. Without
knowing the internal workings of the CP1 effects processor, and without having full measurements of
How would you critique your performance in this setting? What mistakes did you catch yourself
+ Tight
response was the same. That may be an important point: that whether or not a performer perceives an
environment as different, the vocal response may remain the same, if the environment is the same.
This gives credence to the theory that a singer is not necessarily always aware of the changes that an
environment is having on his or her voice. Later, this study presents the opportunity for outsiders to
judge a singer's performance, very much because of the trap of problems with self-judgment.
Standard settings, plus Mackie mixer High EQ all the way down, CP1 ROLLOFF 329Hz. This setting
was designed to contrast with the "High Enhanced" setting, with an attenuated high frequency response.
Acoustically, what would you say this setting lacked? How did you notice it?
RB11: "Consistency in resonance and reverb - sometimes I noticed a good return, sometimes
no return"
RB3: "Lack any sort of resonance or reverb - sound was gone after I sang"
RO4: "I just felt tired and my head hurt -- it was the first time I was aware that I was in a small
room"
- Nothing missing
WM11: "Nothing"
Despite that there was sound coming back to them, the singers did not perceive it, because they were
used to receiving a lot of return at high frequencies. This setting provided feedback only at lower
frequencies. Because of equipment limitations, it was impossible to determine the exact RT frequency
distribution in this setting. The exact line where reverberation becomes "lost" to the performer's ears
RO's comment that her "head hurt" is very interesting. People often have a sensation of cotton-
headedness when they first step into anechoic environments, and this setting was the closest that
52
RO had come to a completely anechoic setting. Although she was getting some feedback from the
room, she was getting a small enough amount that she was starting to feel the effects of being in the
anechoic chamber.
Acoustically, what would you say this setting did not lack? How did you notice it?
RB11: "It didn't lack any sense of resonance -- there was something there."
RB3: "It didn't lack a feeling of space -- I could tell sound was in the room -- it just got soaked
up."
CF4: "It had just enough liveliness to make you hopeful but not enough for you to play with"
- Nothing wrong
WM11: "Nothing"
Apparently this room was a tease, in that it provided some vocal reinforcement, but not enough. RO
commented that it was "even top to bottom," which has been already labeled as desirable in an
environment. All of the above comments seem to indicate that singers would prefer to have a small
It should also be noted at this point that after trial WM10, WM's surveys became less and less
responsive. Further research should include a larger pool of participants and fewer room settings to cut
down on the
+ Hold back
RB11: "I tried to hold back and not overpower the space (not too effectively)"
RB3: "I pulled the sound back -- reduced volume and intensity"
+ Sustain notes to fill the space with voice, rather than relying on reverb to fill it
CF4: "I tried to fill the dead space more and drove myself flat in the process"
This is an interesting case study in musicians' adjustment to an environment that is not very kind to
mistakes. Because there was very little reverberation in the higher ranges, very often the only sound that
the performers were hearing was the sound of their voice as it emanated from their mouths. RB adjusted
by pulling back and cruising through the session. JP and CF held their sound out, trying to simulate the
length of the reverberation that they wanted to hear. RO and WM became preoccupied with the sound of
It may be interesting to follow this route a bit further, and see how singers would react to not hearing
RB3: "Liven it up -- make it more responsive in order to give a better sense of sound in the
room"
CF4: "Liveliness"
This set of data points out that singers do need some feedback from their environment in order to feel
satisfied while performing. Although their terminology differs, all of the singers are essentially asking
Does this setting remind you of any halls you have sung in, or listened to music in, before?
+ Practice/audition rooms
RO4: "Auditions - Auditions!! Which are usually held in tiny dead rooms"
The artists are again commenting that the room seems dead, at least at higher frequencies.
How do you think a member of an audience would rate your performance in this setting?
General impressions were negative, i.e., "dry," "not particularly memorable," etc.
55
How would you critique your performance in this setting? What mistakes did you catch yourself
making? + Tired
CF4: "I was flat; I cracked; I messed up words. You name it, it happened. It was sad."
RB11: "Uneven in places. Different vowel placements and breathing were used to offset
This set of data illustrates a very important point: under some circumstances, despite efforts by singers to
ignore the poor conditions in which they are performing, their voices can be affected in ways of which
they are not even aware. Not a single one of the performers commented on having changed their
technique to a technique that would be tiring, and yet their resulting performances were "tired," or "off
Generally, this setting was not very well liked by the singers. WM was apparently either very happy
with a setting that none of the musicians were happy with (all of the Maxima are hers), or she was not
putting her true feelings on the survey paper. There is also the possibility that the equipment was mis-set
for this particular environment, and the author was not aware of the problem. Regardless, her results can
Standard settings, plus CP1 EFF LEVEL -20dB. This setting was designed to give the singers a feeling
of being in a less responsive hall. That is, although the amount of acoustical feedback was unchanged,
Acoustically, what would you say this setting lacked? How did you notice it?
RB12: "It lacked openness -- the sound never 'filled' the room"
CF12: "It was good in the front -- but not high enough"
RO3: "Not enough resonance. I felt myself pushing and I just couldn't hear much sound"
These results suggest a new hypothesis: singers are very much in tune with the direction from which
their aural support is coming. Three of the singers specifically pointed out that there seemed to be a
disparity between the amount of reinforcement coming from the front versus the back. Independently, it
was mentioned that there was not enough resonance. It is possible that these two phenomena are related.
As soon as they lost support from behind, they lost the sense of "filling" or "resonating" with the room.
Apparently, singers need evenness not only in terms of frequency, but in terms of directionality. At high
volumes, they did not notice as much, but at lower volumes, when the rear speakers may have been
Acoustically, what would you say this setting did not lack? How did you notice it?
RO13: "I'm not sure - it had an evenness of acoustics from top to bottom"
+ Some presence
RB12: "It didn't lack resonance -- the sound was present in the room -- it just didn't move"
Again, it is shown that singers prefer evenness across frequency range to just about any other feature.
CF and RB acknowledged that there was some support, but there was not enough, suggesting that
although
58
room acoustics should be subtle, they must be loud enough to be clearly audible.
How do you think you adjusted, in order to sound your best in this environment?
RO13: "I had to ... lower the breath. I also probably pulled back a bit"
RB12: "I tried not to let the sound spread -- I tried to keep it focused"
Interestingly, the singers felt constrained by the environment. Even though it was exactly the same as
the first environment, at a lower volume level, they were more consciously aware of the environment's
limitations. It is worth pointing out that the singers seemed to seek out the source of their most support,
and homed in on it: RB and WM both speak of "focusing" and "singing forward," where they perceived
RB's comment could also be considered supportive of RO's comments, in that "focusing" could also
mean pulling the sound back and concentrating on filling a smaller space. In performance conditions,
If you could change one specific acoustical property of this environment to make it better, what would it
+ More support
RB12: "I would give it more space. It is important for a singer to have the feeling of sound
59
filling a space"
RO3: "It needs more resonance -- so I would push (and eventually trash my voice)"
Singers apparently need to be fully enveloped, acoustically, in order to get a sense of an environment
supporting them. Otherwise, the result can be, as RO says, "trashing" of the voice.
How do you think a member of an audience would rate your performance in this setting?
Standard settings, plus CP1 SYSTEM/EFFECT MUTE. This setting was simply the "bare" sound of the
Acoustically, what would you say this setting lacked? How did you notice it?
JP5: "Everything"
Several of the singers described the sensation of singing in an anechoic environment as having their
performance "swallowed" by the room. This is the only room in which the singers were unanimously
Acoustically, what would you say this setting did not lack? How did you notice it?
The singers were again unanimous, in observing that the setting lacked everything. WM phrased it
interestingly: "when I stopped singing no sound continued in the air." This hints at the idea that singers
may not ever be aware of the individual wall reflections that affect them; that they may have a much
perhaps singers do not need a lot of feedback, so much as they need consistency across direction and
frequency.
How do you think you adjusted, in order to sound your best in this environment?
CF8: "I didn't this time ... I just rolled with it"
- Didn't adjust
- Increased
A majority of the singers seem to reach a point where they are aware of some sort of vocal peril that is
imposed by the environment in which they are singing. A good topic for further study would be
determining at what point singers decide that the environment is hostile, and to simply "roll with it." It
is a safe assumption that the acoustics in some halls fall right on the dividing line, in which singers are
unaware that there is anything wrong with the acoustical support their environment is providing, but in
CF also points out that she held her phrases for longer periods, again trying to create artificial
reverberation by simply extending her notes. She also points out that that technique makes expression
62
very difficult. However, she changed her technique in CF8.
If you could change one specific property of this environment to make it better, what would it be? Why
+ Liveliness
+ Reverb
WM1: "Reverb"
Obviously the anechoic environment is an extreme case in room hostility. The author originally was not
going to include it, but upon further thought, it seemed wise to include a logical extreme in the changes
in parameters--a reference point, or acoustical "absolute zero," to find out the effects of no environment
on singers.
This points out an interesting angle for more research. This study is subtractive in nature. That is, it
begins with a "standard environment" that has some presence, and certain characteristics are removed to
produce an effect. It may be more revealing to start with an anechoic environment, taking an additive
approach, as it might provide more insight into what singers really need from environments, and would
give a good idea of the dividing line between the realization of danger and the range of vocal peril that
for acoustic absolute zero, it might be a good idea to begin every session in this environment, have the
performers begin singing, then to turn on the actual artificial environment. This would allow a
"resetting" of the singer's sense of environment, rather like drinking water between wine tastings.
Does this setting remind you of any halls you have sung in, or listened to music in, before?
WM10: "closet"
How do you think a member of an audience would rate your performance in this setting?
Responses were generally not positive. RB says, "Yuck! No movement or energy." That just about
How would you critique your performance in this setting? What mistakes did you catch yourself
+ Forgot words
WM10: "I forgot some of the words in 'Star Spangled Banner' having to scream"
WM1: "forgot words" [WM had sung the SSB more than 40 times before]
+ Pitch problems
JP5: "I found the pitch was harder to maintain. I sometimes felt the pitch was about to change"
64
WM seems to have a problem with performance nerves. In really poor environments (consistently), she
forgot the words to the Star Spangled Banner. Apparently acoustics can throw off more than just the
sound of a performance.
More interesting is the agreement between CF and JP that their pitch suffered in the environment. This
suggests that singers may use reverberation as a sort of "pitch memory," in the same way that people use
environment, they use that returning sound to launch them into their next pitch. This very strongly
suggests an experiment with very subtle pitch changes in the reverberation of an artificial environment.
described the environment as being completely non-reverberant and non-resonant, they still found it
responsive--suggesting that, as long as they can hear the sound of their voices coming directly from their
Standard settings, plus CP1 SHAPE Fan. Understanding the SHAPE parameter of the CP1 effects
processor requires a brief explanation of the way the CP1 operates, internally. When the CP1 receives a
sound at its inputs, it divides its response to the sound into two pieces: an early reflections stage and a
reverberant "tail." The early reflections stage applies to the first few dozens of milliseconds of
reverberation, and consists of many distinct repetitions of the original sound--representing reflections
from distinct surfaces--at varying low volume levels, sent to the six different output channels. This
approximates the response of a real room to the sound. At the end of the early reflections stage, the CP1
smoothly transitions into the reverberant tail, which is more diffuse. The SHAPE parameter affects only
the direction and number of reflections returned by the CP1 during the early reflection stage. The
Acoustically, what would you say this setting lacked? How did you notice it?
+ Headroom
JP3: "Responsiveness"
Overall, the singers seemed to think that the FAN SHAPED setting was much more natural, and more
friendly, than the RECTANGULAR setting. The study was originally supposed to have the CP1 set to
FAN all the time, but a clerical error with RB's first trial made that impossible -- the CP1 had reset to
RECTANGLE, and the author was faced with either having to start RB back at the beginning of the
The "perfection" of the setting described above seems to be completely divided along the line of gender.
It would have been desirable to have more subjects, so that true statistical significance could have been
established regarding the preferences of singers of certain ranges for certain parameter settings.
Acoustically, what would you say this setting did not lack? How did you notice it?
+ Reverberation
JP3: "Reverberation"
RB7: "Reverb"
CF2: "reverb. I could hold notes as long as I wanted without the room dictating it"
RO notes that this environment had "plenty of room without being too live." This verifies the findings
of the many studies that have determined the importance of the placement of early reflections on the
desirability of certain settings. This is another topic area that could be expanded into an entire study of
its own.
67
How do you think you adjusted, in order to sound your best in this environment?
CF2: "I was more dynamic and let the room carry the sound"
Other comments were interesting, including JP's idea of "concentration, fall back on feeling," and WM's
If you could change one specific acoustical property of this environment to make it better, what would it
+ Don't change it
There was no meta-statistically significant correction suggestion, which seems to indicate that this is the
most desirable environment of the ones used in this survey. One likely explanation: the FAN shape is
more natural to a singer's ears than a RECTANGULAR one. This may deal with the direction, the
number, or the spacing of early reflections. Although WM mentions a "dead space" to her sides, the
author is unsure what the difference is between the early reflection patterns produced by
RECTANGULAR and FAN. The only definitive conclusion that can be drawn is that something about
+ Cathedral
RO7: "Large stone Catholic churches as opposed to Protestant wooden churches. Cathedrals
The significance of RO and RB agreeing on an exact setting can not be overemphasized. Not only do
they both agree that this setting has stone walls, they agree that it has high ceilings! This specificity
suggests that this is indeed the most realistic environment, and it could very well be that some vertical
directionality has found its way into the early reflections--that is, a sonic illusion that the singers are
This brings out two ideas: first, further study should be done into the differences between the FAN and
RECTANGULAR settings on the CP1. Second, and more importantly, future studies like this one,
which use artificial environments, should include the third dimension when studying directionality
preferences.
How do you think a member of an audience would rate your performance in this setting?
All of the singers had positive things to say about an audience's response, except for JP, who remarked
that they would rate him as "strained." Perhaps he was the only user of the environment to reach
How would you critique your performance in this setting? What mistakes did you catch yourself
WM12: "my voice was a little weak, but this acoustic environment wasn't bad"
69
CF13: "Tired. Yukky. Not good quality at all"
Could this environment be deceptively good? It seems almost as if, in the tenors' low ranges, the
environment holds much promise, but as soon as they begin to sing higher, they come up against a brick
This is a good point to note that seven trials (which translates to 21 songs) seems to be the point at
which singers become uncomfortable with singing any further. This could explain the "tiredness" factor
in all cases except JP3. However, it presents the point that future studies should reduce the amount of
time that singers spend singing straight through, and perhaps should concentrate on finding a larger
Again, the statistical significance of these numbers is questioned, for several reasons: first, certain
environments were rated twice by performers who tended to rate settings high, and others were rated
twice by performers who tended to rate settings low. Some sort of statistical adjustment might make
these numbers more informative. Second, it is possible that, as singers discovered what the truly bad
environments sounded like, they tended to overrate the good ones, meaning there is some time-variance
in their rating scheme. This should be somehow planned or in future studies--perhaps by not doing any
numerical study at all, or by making a more fixed set of criteria for rating the settings.
Standard settings, plus Side amplifier (MT600) turned off. This setting was designed to have less lateral
energy than the other settings, to test the effects of directionality on singers' response.
Acoustically, what would you say this setting lacked? How did you notice it?
+ Low response
WM5: "When I used my lower range, I could not hear any reverb"
+ Reverberation
RO8: "I think it lacked some reverb. I don't think I would have sung as big if it didn't. I felt my
71
voice getting trashed in the upper notes of my selection"
RO2: "I felt like the top notes didn't have enough room to bloom"
RB13: "The setting lacked expansiveness -- the sound did not blossom"
Again, it is interesting to find that RO and RB used the same terminology without realizing it. Although
missing low support was a running theme among all of the environments, "blooming" and "blossoming"
seem to have become an issue only when the side speakers have been cut off.
Because theatrical and musical stages are usually less deep than the hall they open into, it is possible that
singers are used to hearing most of their early returns from in front of them. The idea of "blooming"
could be explained by the side returns becoming larger over time as the singer "fills" the hall.
Acoustically, what would you say this setting did not lack? How did you notice it?
+ Reverberation, resonance
RO8: "It [had] wonderful fast resonance. I could sing big without pushing"
The singers seem to be contradicting themselves, as to whether there is enough reverberation present in
this setting.
72
Off the record, several of the singers made comments about this setting that led the author to believe that
too much return directly to the sides of a singer can hamper a singer's performance. Interestingly, this is
the opposite of what audiences prefer, as seen by Dolby Laboratories' success with Surround Sound,
which includes the addition of speakers to the sides of the audience to envelop the listener in the sound
It would seem that singers are not comfortable being enveloped by their own voices, while audiences are
very happy to be enveloped. Although RB's comment about needing more "room" could very well be a
request for more side returns. This particularly reinforces the statement that was made earlier about
How do you think you adjusted, in order to sound your best in this environment?
+ Don't adjust
+ Avoid Darkening
WM5: "I had to sing with a medium force in my lower range, so it didn't sound gutteral"
- Darken
All of the above comments seem to indicate individual preference, rather than an overall trend toward a
+ More reverb
WM5: "Add a slight bit more reverb. I sounded throaty in the lower ranges"
RO8: "For art song + some oratorio I would ... put in some reverb"
Three answers which are worth comment, which were not mentioned by more than one performer, come
from RB and RO. RB restates that he would "give it more room." RO mentioned that she thought this
setting would be perfect for opera. This may be because of the 'stage' effects mentioned above. RO also
brings up a new point, saying "I would enhance whatever makes it possible for the singer to hear
himself/herself as the audience might be hearing it." RO had commented previously that this setting
seemed to give her a "little bit of what I must sound like to an audience listening." This could be a
function of directionality, or could even be a fluke, but the underlying idea--that singers would like to be
able to hear themselves as the audience is hearing them, and thus enhance their ability to give a near-
The most interesting results from this iteration were not numerical, but were qualitative, stemming from
RO's comments about this environment being "perfect for opera." Her comment reiterates what all
preliminary literature searches on ideal acoustical conditions revealed: different genres of music tend to
have different acoustical needs. Ideally, then, singers should always sing in adjustable environments, so
that they can tailor their environment not only to their needs, but to the intrinsic needs of the genre of the
Environment 9: "Reversed"
Standard settings, plus Main channels feeding rear speakers, Rear channels feeding front speakers. This
Acoustically, what would you say this setting lacked? How did you notice it?
+ Resonance
RO10: "It needed a little more resonance for the high notes"
75
RB and RO are noticing that they are no longer getting non-delayed, direct feedback of their voices from
in front of them. Although this feature (which is an artifact of the CP1's design for use with home
theater systems) seemed artificial to them at first, now they seem to miss it, and associate it with
resonance.
Acoustically, what would you say this setting did not lack? How did you notice it?
+ Surround Sound
RB9: "I was noticing sound from different areas (side and rear)"
Again, the singers have fallen into stressing the point that a performance can be affected by
directionality of acoustical feedback: if it was worth mentioning, then it must have an effect.
How do you think you adjusted, in order to sound your best in this environment?
+ "Easy?"
CF5: "I relaxed and took it easy. The room could do some of the work"
It is possible that, because the majority of the singer's voice (the non-processed, non-delayed return) is
coming from behind the singer's head, CF and RO are perceiving it as part of their own vocal
production. This leads to an interesting question: should vocal reinforcement systems, which are used to
amplify a singer's voice in large halls so that all members of the audience perceive it at acceptable
volume levels, broadcast their reinforcing signals behind the head of the singer they are reinforcing?
CF3: "OK -- my throat is tired. I'm cracking more and going flat on the top notes"
RB9: "Tight & tired -- clutches and grabbing due to tiredness" [RB also mentions, in response to
How do you think you adjusted: "I was aware of intonation problems -- (perhaps setting,
perhaps tired voice) and tried to adjust by not listened too closely to the sound"
For some reason, this environment seems to be consistently producing intonation problems. Despite that
it is exactly the same as setting one, simply reversed, the reversal has really caused troubles for the
singers. This could be the result of several phenomena: perhaps the singers have grown so used to the
solid frontal feedback, that they can not keep themselves in tune with less. Another theory, based on
research which shows that pitch perception changes with volume, is that the high volume of their
Standard settings, plus Front amplifier (D-75) and rear amplifier (MT2400) turned off. This setting, like
Acoustically, what would you say this setting lacked? How did you notice it?
JP1: "Overall, it lacked feedback. I was missing much of the richness of the tone"
RB10: "It lacked a sense of forward sound -- ... there was little resonance or reverb"
CF9: "Front sounds. I heard a lot from sides and not much from the front at all. It wasn't
All of the singers noticed that something was missing, but only two realized that it was a lack of sound
coming from the front and rear. In comparing this to the "no sides setting," it becomes apparent that
1
2
although feedback from the sides is desirable, singers do not miss it if it is gone. Front/rear feedback is
How do you think you adjusted, in order to sound your best in this environment?
+ Sing lighter
RO5: "I sang much lighter. I really had to pull back on the upper range"
- Sing louder
WM6: "I feel I had to sing much louder and stronger to be heard"
JP1: "My technique and schooling rely on not adjusting to feedback but relying on feeling alone.
If I did react I would think I would try to push more air, more volume. But I try not to."
RB10: "I tried not to fight the room -- allowing the sound to go where it would"
The singers are genuinely having problems with the uneven directionality of the room. Without more
research, an absolute determination of the importance of each direction can not be made.
If you could change one specific acoustical property of this environment to make it better. . .
+ More reverb
RO5: "I think more reverberation. I could use more color in my voice"
Although the singers are voicing their desires differently, they all seem to be voicing the same idea: they
want frontal reverberation. JP's comment is particularly interesting because he does not seem to know
Does this setting remind you of any halls you have sung in, or listened to music in, before?
JP1: "It has a certain 'deadness' that feels like an open space or a pillow in front"
How do you think a member of an audience would rate your performance in this setting?
RB10: "Where did the sound go? It never really reached me ... "
CF9: "It felt like there was nothing in front of me -- would the audience even hear me?"
JP1: "If they are in a direct line they might not think it too bad"
These comments echo RO's desire to hear exactly what the audience hears. They reveal a singer's
mistaken impression that if (s)he hears something on the performance platform, the audience is hearing
the same thing. Previous research has shown that this is absolutely untrue, although singers would love
WM6: "Horrible -- I couldn't get into singing -- I became careless about the quality of my voice"
All of these answers stress the importance of frontal feedback for a singer. They also suggest that a
singer's energy level in performance may be affected by the room's return, in the same way that the
audience is emotionally affected by the singer's performance. A room may act as not only as a "pitch
This section explores the correlation between the opinion of a singer for his own performance, and the
impressions of impartial listeners. Recordings of the singer's performances were played back through a
high fidelity system to a panel of three musicians. The recordings were unprocessed--that is, the
reverberation effects produced by the CP1 in the original experiment were not recorded directly. The
listeners were asked to fill out a questionnaire about each performance while listening (which can be
found in Appendix 8). This section presents a comparison between the impressions of the musicians,
which are described in the previous section, and the impressions of the listeners as written on the survey
sheets.
The method of analysis for this section is very different from that of the previous section. It is an
unfortunate fact that the questions the performers answered regarding their performance were not the
same as the questions which were given to the impartial observers. Because of this, objective, statistical
comparison of the singers' impressions to the listeners' impressions is almost impossible. Instead, a
Each singer's data is presented alone, sorted by room setting. The letter/number code in the header for
each section, of the form XX#(#), denotes the singer's initials, the trial number for that particular singer,
Presenting all of the data from the singer's survey sheet would be redundant, because that data was
already presented in Section One. Instead, focus is placed upon the two most relevant questions to this
section: "How do you think you adjusted, in order to sound your best in this environment," "How do you
think a member of an audience would rate your performance in this setting," and "How would you
6
critique your performance in this setting." Along with the answers to these questions, each section lists
the non-numerical impressions of the listeners, in response to the question "What vocal adjustments /
performance characteristics make this performance different from all or some of the other performances?
Once the answers to these questions have been presented for all thirteen trials, all of the numerical data
from the singer's survey sheet is presented in a graph, along with a graph of the numerical data from the
listeners' survey sheets. Note that the singer's impressions have been inverted--that is, the results have
been subtracted from 100 to yield a "best-to-worst" graph, rather than a "worst-to-best" graph. A
comparison of these two sets of graphs, combined with the non-numerical data, will verify whether the
singer's impressions of his or her performance correlate with the impressions of the objective set of
listeners.
Because of time limitations, it was only possible for the impartial listeners analyze two of the five sets of
performances: RB and RO. WM's performances were analyzed up to WM3, so the data has been
analyzed up to that point, and the remaining ten performances have been left blank.
7
Performer 1: RB
Forced
When I noticed a decrease of reverb or resonance in the upper range, I tried to back off the sound
's'es better
Very good
I tried to let the voice fill out -- both top and bottom
Gasps
Happy
A lot better
Oversinging (dramatic?!)
Perhaps I oversang -- let the voice go too far -- breathing was inconsistent in Amor Ti Vieta
Smoother
Very trebly
Tired
I tried to fill out the bottom a bit more to compensate for the brightness in the top
Tired and tight in the top -- changed breath spot in SSB [ed. Star Spangled Banner] and couldn't
Very tired
Tired
I tried to hold back and not over-power the space (not too effectively)
Uneven in places--different vowel placements and breathing were used to offset tiredness
in the voice
10
RB3(4) Listeners' Impressions for "High Inhibited"
Overall depressed/depressing
Generally OK
Very tired
Fatigue
Uneven -- bland
Inconsistent -- vowels and breathing are not consistent due to vocal wear and tear
Better pitch
Relaxed
Generally Even
Lackluster -- tired -- tight in the top, glottal clutch at the end of Amor Ti Vieta
Tight in the top -- voice is getting tired. A few clutches from time to time. Breathing
is inconsistent.
Tired.
Tired and inconsistent. Top is weak and thin. High note in Amor Ti Vieta was strained
and forced.
Trebly
Too bright
I was aware of intonation problems (perhaps setting, perhaps tired voice) and tried to adjust
Inconsistent
I tried not to fight the room -- allowing the sound to go where it would
20
1 5
10
20
15
10
5
30
20
10
15
10
15
10
100
50
Graph 6: RB's Opinions [100% - X] (Darker: "How Powerful..." Lighter: "Overall") Ana
lysis
16
First, it should be noted that overall, RB had more vocal glitches than any other kind. This is probably
because the "vocal glitches" category contains the largest number of possible problems, including vocal
There is a variety of ways that the above graphs could be interpreted. This paper summarizes the set of
performances in terms of whether the evaluator graphs (graph 1 to graph 5) follow the same contour as
the singer graph (graph 6), and points out the sessions for which the contours vary. This comparison can
Generally, a glance at the Table 1 shows that RB's impressions of his performances follow the same
contour as the impressions of the listeners, although the scale of the changes is exaggerated. This is due
to discrepancies between the structure of the singer surveys and the evaluator surveys, which is
In some cases, the contours of the two sets of graphs differ greatly. In general, RB's performance in
setting 6 ("Anechoic") was equally as good as his performance in setting 7 ("Fan Room"). Yet he felt
much better about his performance in 7 than in 6. He also thought set 6 was worse than set 5 ("Less
Presence"), when in fact the opposite is true. His mistaken impression about performance 7 may arise
from a "rebound" effect, or a period of recovery, from setting 6. This suggests that previous settings
may have a profound effect on the perception of current performance. He also did not feel very good
about his performance in setting number 10 ("Sides Only"), which was rated quite well by the objective
listeners. Perhaps the unnaturalness of having sound returning only from his sides made him uneasy. It
is also possible that fatigue played a factor in his impressions of his performance.
17
Did much Did worse Follows Did better than Did much
worse than than thought contour thought better than
thought thought
"Standard" VBPES
"High Enhanced" B VPS
"Most Live" VBPES
"High Inhibited" VBPES
"Less Presence" VS BPE
"Anechoic" V B PES
"Fan Room" V BPES
"No Sides" B E VPS
"Reversed" VBPES
"Sides Only" VBPES
Voice doesn't have enough room particularly in lower range -- sounds thin + dry
Too fast
Blah
Mediocre. I didn't really think about breath support. And I could have been more musical--
A bit oversung
More legato
Sound tired
Harsh "th"
The "America the Beautiful" was better than the Star Spangled Banner because I darkened the
sound but I think both were very good. The art song was decent.
Pretty good.
20
RO4(4) Listeners' Impressions for "High Inhibited"
Stuck just low of placement but very even quality all the way through
Too Italian
Piano -- tinny
I had to darken the sound and lower the breath. I also probably pulled back a bit.
Pretty good.
A little pushed.
Notice "scooping"
Lovely sounding until gutteral notes at the end [of chosen piece]
Obvious tried to brighten to keep pitch up -- but because not high enough placement pitch is still
not quite on
Stuck in low vocal position / only burst of energy gets it a little bit into correct position.
Pretty good.
Consonants better
I didn't have to [adjust] too much -- I had to be careful not to push but I think that was more the
This was much more even top to bottom with regard to timbre + quality.
Clean sound
Good but better when I pretended that I was singing opera -- especially in America the Beautiful.
Pretty good.
23
I just think I could do everything more especially sing with longer line.
Sounding tired.
Present
More presence
I sang much lighter I really had to pull back on the upper range.
It would depend on the audience. A real opera audience would probably boo you off the stage.
Whereas an "earlier than thou" pretentious recital audience would probably find it
delightfully "safe."
30
20
10
15
10
20
10
15
10
15
10
100
80
60
40
20
Graph 12: RO's Opinions [100% - X] (Darker: "How Powerful..." Lighter: "Overall")
Analysis
As with RB, the comparison of the contours of RO's opinion graph and the objective listeners' opinion
Overall, RO's impressions of her performances follow very closely with the opinions of the objective
listeners. This is especially pronounced with regard to setting number 7, which she rates the most highly
of the ten rooms. The evaluators agree that her best performance was produced in that environment.
27
Surprisingly, RO enjoyed setting number 6, which was supposed to be the worst setting for a singer to
perform in. The only area in which she did exceptionally poorly in setting number 6 was pitch, which
further suggests that singers use the concert hall as a "sounding board" to maintain their tuning and key.
Did much Did worse Follows Did better than Did much
worse than than thought contour thought better than
thought thought
"Standard" VBPES
"High Enhanced" V BPES
"Most Live" VBS PE
"High Inhibited" VBPES
"Less Presence" VBPES
"Anechoic" P VBES
"Fan Room" E VBPS
"No Sides" BPES V
"Reversed" VBPES
"Sides Only" PE VBS
Choppy beginning
Better resonance
Better presence
Consonants
Just sang and tried not to listen because it was too confusing
This was much cleaner and not breathy like the other two times
Tired / PRESENT
Consonants
This is the type of setting I am used to and does not take much adjusting
Pretty good
I felt fairly satisfied -- a few glitches. I feel comfortable in this setting because I could play my
Consonants
Listened to the sound of my voice coming from me, not off the walls -- of which I'm not used to
Forgot words. Since my sound ends once I stop singing, I needed to breath [sic] quickly so no
30
dead space would occur.
20
15
10
5
15
10
15
10
20
15
10
5
15
10
150
100
50
Analysis
Because the objective listeners never finished listening to WM's performance set, it would be useless to
present her results in the form of a graph. However, there are some features of the limited amount of
similar performances in settings 3 and 6, and a better performance in setting number 5. WM's opinions
of the settings does not reflect this at all. Instead, WM seems to feel equally comfortable in settings 3
and 5, and uncomfortable with setting 6. It would seem, based on the small amount of data available,
that WM's opinion about an environment has little effect on her performance quality. The reverse is also
true.
It is possible that, because of WM's lesser experience when compared to RO and RB, she is not as aware
of her level of performance quality. This could only be determined through further study.
34
Conclusion-
Although there were too few participants in this survey to establish any statistically definitive answers to
the questions posed in the introduction, several interesting points did come up in the course of the
analysis of the data gathered which are worth noting, and which provide a solid foundation for future
research. Although this research did not provide any statistically significant answers about the effects of
concert hall acoustics on performance, it provides some very specific questions. That is, hypotheses
The theme which recurs most often in the gathered data is the desire by musicians to perform in even
environments. Performers prefer, and actively ask for, evenness in frequency response and sound
distribution. When faced with an uneven performance environment, the singer will adjust the singing
technique to force the environment's response to be more even. In environments with frequency
problems, the singer may darken or brighten the vocal timbre. To fill directional holes, the singer may
turn his head while singing. In non-reverberant environments, the singer will hold notes for a longer
period of time.
singer's sound, such as breathing or vocal stops. In these cases, there is very little that the singer can do
to correct the problem. For example, a singer certainly can not stop breathing. However, despite that
logical assertion, the singer may actually unconsciously try to stop breathing, or at least to quiet the
Most often, a singer's adjustment is completely unconscious, and is a natural reaction, even when the
singer is not aware that adjustment is occurring. In some cases, the singer can become fatigued, sore, or
35
unmotivated without even realizing that the room is the cause. This is illustrated particularly well by
RB's situation in the "Most Live" setting, in which he responded verbally differently in his two
evaluations, but in which he responded physically identically. The voice reacts the same to an
There is a point, however, at which the adjustment has become so extreme that the singer can not help
but realize that it is occurring. When this "peril point" has been crossed, the singer will consciously
make an effort not to adjust. The consciousness of having crossed this point makes the difference
between a singer holding back and adjusting to an environment, or singing all-out and refusing to adjust.
The peril point seems to differ from singer to singer, and seems to vary directly with the singer's level of
solo experience.
Despite their listening very carefully to their performances, sometimes singers are deceived into thinking
that they are delivering a poor performance, even when the performance is flawless. Although singers
are aware that the acoustics on the stage are not the same as the acoustics in the audience, it is
impossible for them to judge their performance any way other than with their own ears. Oddly, though,
the participants in this experiment sometimes perceived the same environment very differently in
different trials. It is likely that these differences in perception were purely contextual. The singers' ears
had adjusted to the sound of the previous environment, and that memory colored their perception of the
present environment.
This phenomenon is further supported by the fact that, toward the end of their trials, the singers seemed
perfectly at ease in the artificial environment that they had previously thought sounded treble-heavy and
unnatural.
Although some environments can apparently be made to sound better by comparison to others, the vocal
response of the singer has nothing to do with how the environment is perceived at the time. In other
36
words, the physical response of the voice has little to do with how the singer may feel about the
environment, but has a lot to do with how the singer feels in the environment. A singer may think that a
setting is relatively good, and yet may not sing well because he or she is not aware that the room is
The most interesting effect observed in this particular run of settings was the "fight or flight" reaction of
RB and CF in one of the particularly loud environments, which drove them to sing their voices raw.
Although the adrenaline rush resulting from having one's voice amplified and returned directly in front
of one's face can be quite exhilarating, having an excessive energy level does not make for the best
performances, and can be quite exhausting over time. Fortunately, in real-life performing environments,
the volume of the return of the hall never reaches the "fight or flight" volume level.
Some interesting results came from the use of the anechoic chamber's natural acoustics. Although some
of the performers commented on how even the acoustics of the chamber were, no performer was
completely happy with its response to their voices. Their pleasure at the chamber's natural acoustics
only furthers the hypothesis that ears can be trained to appreciate any environment, even one as
unfriendly as an anechoic chamber. It was established, with all of the singers, that although performing
with no return from a room is not pleasurable, it is possible. And it is interesting to note that all of the
singers were consciously resisting the urge to adjust to the anechoic chamber, which supports the
hypothesis that there exists a point where a singer will become conscious of vocal peril, and re-adjust.
Another theory which came about because of the use of the anechoic chamber is the idea of the use of
reverberation as a sort of "pitch memory." In much the same way as human beings use repetition to
remember a series of numbers, musicians use a hall's natural repetition--reverberation--to hold their
pitch.
37
In environments which lack reverberation in crucial frequency ranges, singers tend to have more
troubles holding on to the pitch, and tend to perceive themselves as falling out of tune. They may adjust
by holding their notes for a longer period of time. However, this adjustment causes problems with
Regarding sound distribution, the singers had definitive preferences for the direction of the return of
their voice from the environment. In this six-speaker setup, the front is without a doubt the most
important source of a singer's feedback (which is logical, since human beings are visually-frontal beings,
and usually associate sounds with visual cues). Secondarily important is feedback from the rear. The
sides seem to be almost an afterthought, which is not consciously missed when it is not present, but
Section Two of the Analysis portion of this paper demonstrates that singers' impressions of their
performances are, for the most part, quite accurate when compared with the impressions of objective
listeners. There are some circumstances, though, under which a performer may judge their performance
as much better or much worse than it really is. These circumstances seem to involve context more than
absolute room acoustics. For example, Room 7 seemed to be the favorite of all of the singers, yet the
performances in Room 7 were not judged as being particularly better than any of the other
performances. It is likely that because the singers were the most satisfied by Room 7, they thought their
performances would be perceived as better. The exact opposite case holds for Room 6, in which the
singers did not feel powerful or effective at all, and yet in which they sometimes delivered performances
which were comparable to their performances in Room 7. These two rooms--supposedly the best and
the worst in the set--were contextually very different from the other settings. Their difference from the
other settings were the most dramatic, and therefore the singers felt the most affected by them.
38
Fatigue also played a very important factor in the singers' impressions of their performances. A good
example is RB, who thought that his performance in Room 10, one of the last four of his performances,
was terrible. In fact, his performance was no worse than any other. Yet RB seemed to have a very
strong need for a return of his voice from the sides, while the other participants did not.
Generally, then, singers usually have a very accurate idea of the quality of the performance they are
delivering. Each singer has a preference for certain settings, and will judge their performance based on
their comfortability level with that setting. In circumstances where the singer feels terribly about the
performance setting, the singer will almost inevitably feel terribly about the performance, even if it is
stellar.
All of these hypotheses could benefit from further experimentation to determine their validity. The
following section, written for the enterprising acoustician, describes the problems encountered in
performing this experiment, and contains suggestions on conduction of future research in this field.
39
For Future Research-
The driving premise behind this project was that, regardless of the data acquired, the project report
should be made into a guide for researchers conducting this type of survey in the future. This section
summarizes all of the problems, both expected and unexpected, that were encountered during this
research. It contains suggestions for new experimental procedures which may help to avoid these
problems, and which could result in more conclusive data than is presented here. Suggestions for new
research topics are also presented. It concludes with a note about budgets and funding. It is the author's
hope that this work will inspire larger, more well-funded, more conclusive efforts in this field and
related fields.
EXPERIMENTAL METHOD
In conducting the experiment, a number of problems were found with the equipment and the
experimental procedure which could not be fixed at the time. However, a revised experimental
procedure should take all of these problems into consideration. The result would be a much more
Subjects
The decision to use singers in this experiment was explained in the body of the report. However, there
are some additional notes about the use of singers which were not evident until the experiment was
already underway. These points are important to any future research in this area.
The first obvious problem inherent in using singers was the need for accompaniment. Although this was
overcome with prerecorded tapes, the accompaniment playback volume was inconsistent between
40
sessions, because each singer had their own accompanist record their tape. Instrumentalists would have
no such problems.
Fatigue was also a factor in the singers' performance over time. Pianists have much more endurance,
because their performance involves mostly their hands and feet. Vocalists are severely limited by the
Although these factors could be lessened by doing fewer trials on each performer, the warmup/fatigue
curve would inevitably affects the overall performance quality over time. Although it would be
desirable to account for this in the data analysis, with the information currently available in the
In future experiments, then, the number of performance sets should be restricted to six or seven, or the
number of songs in each set should be reduced. Unless the difference between a cappella singing and
accompanied singing is specifically being studied, the number of songs in a set could actually be cut to
one. In general, singers seem to be capable of approximately one hour of singing, plus brief breaks,
before fatigue begins to affect mental composure and performance quality. Total warm-up usually
occurs after the first twenty to thirty minutes of singing. This leaves approximately thirty to forty
minutes of rock-solid, consistent singing time during which the performer is completely warmed up, but
not fatigued.
Equipment
A properly designed equipment setup can make a profound difference in the reliability of data, the
comfort of the experimenter, and the flexibility of the experimental procedure. All of the equipment
41
used in this project was borrowed, meaning it was impossible to plan the environment from scratch.
Although
42
the equipment used in this experiment was of exceptional quality for the applications for which it was
designed, it fell short of creating a flexible, realistic artificial acoustical environment. Assuming that
future experimenters will have the luxury of choosing their own equipment, there are several equipment
The microphone/speaker/time invariant DSP processor combination resulted in the type of high-
frequency enhancement that occurs when a PA system is on the edge of acoustic feedback. This caused
the singers to be quite irritated by the environments that had a lot of high frequency content by
definition, and caused every environment to sound a bit unnatural. Placing a high-impact microphone
on a boom in front of the singer's mouth would ease this problem. Such a setup would also make it
A lot of equipment was needed to create the environment, and the setup had to be reassembled from
scratch for each singer. Although every care was taken to make each setup exactly the same, it was
impossible to guarantee that the equipment was set exactly the same every time it was set up. Because
the effect is random, it would be impossible to account for it in the data analysis. A permanent setup
would be desirable.
The only frequency controls in this system were the EQ controls on the mackie mixer which was being
used as a microphone preamplifier, and the internal DSP controls of the CP1. A 30-band graphic
equalizer should be included in any system design to make frequency effect tests more subtle.
The CP1 DSP processor was designed for home theater and hi fi systems, not as an artificial acoustical
environment processor. Because of this, it feeds its inputs directly to some of its outputs, and mixes the
43
processed signal on top of the direct signal. This resulted in an instant "slapback" of the singer's voice,
which could not be avoided, and also contributed to the acoustic feedback problem described above.
Further, its controls are all calibrated in terms of home theater and hi fi audio parameters, not room
acoustical parameters, which made it difficult to vary the artificial environments in terms of Gade's
measurement system. Use of a more appropriate DSP system is the most crucial change described here.
One piece of equipment which could be explored is a real-time auralizer. As the cost of these units
drops, it is becoming more feasible to record binaural room impulse responses with a dummy head, and
to create a "room" using a set of headphones. Because binaural techniques have trouble placing
reflections in specific locations, this would make it impossible to test the effects of directionality.
However, a system like this would be much more convenient to transport and set up than a set of
speakers and amplifiers in an anechoic chamber. For experiments that are not concerned with
directionality effects, a real-time auralizer and a pair of good closed-air headphones would be a
reasonable alternative.
If directionality is a concern, with a bit of a budget it would be possible to use a Lexicon Lares system,
set up in a Wenger Acoustic Variable Environment (WAVE) Chamber. These environments are designed
to produce extremely realistic acoustical environments, and although WAVE chambers are designed to
be set up permanently, the components used in its construction are capable of parametrically altering the
simulated environment. Such a chamber is additionally advantageous because it has already been
designed. The user need only call Wenger and order one.
44
Procedure
Because consistency was a very important part of gathering reliable data, it was impossible to alter the
experimental procedure once the project had begun. In retrospect, there are several techniques that
Although it is beyond the scope of this report, this research suggests that the context in which a singer
encounters a particular environment affects the singer's perception of that environment. For example, it
seems that singers are relieved upon leaving the anechoic setting (setting 6), and perform extremely well
in the environment that follows it. Although contextual effects are a fascinating field for future research,
in experiments like this one they are undesirable, and steps should be taken to reduce their effect on the
data. Sheer numbers of singers would solve this problem, if each singer encountered the settings in a
different order. Alternatively, each performance session should be separated from the previous one with
a buffer period in an anechoic setting (or some other neutral environment) for the singer to warm up and
Also beyond the scope of this report is the phenomenon of ear adjustment and coloration. As the
experiment progressed, the singers became more accustomed to singing in artificial environments. What
seemed artificial to them at the beginning of the experiment did not seem artificial by the fourth or fifth
performance session. This could have been corrected in one of two ways: either the first two
environments could have "warm-up rooms," discarded during the analysis, or non-artificial-sounding
Also, it was briefly mentioned in the data analysis section that this survey is "subtractive" in nature. The
standard room from which every other room is derived, has some ambience. It might be wise, in future
45
experiments, to use the anechoic chamber as the starting point, and to base all of the other rooms off of
that, adding one bit of ambience at a time, carefully choosing the environments such that a single
acoustical feature is added in each one. Assuming that the results could be superimposed, this would
provide much insight into the reactions of singers to very specific forms of a hall's acoustic feedback.
Finally, future experiments should make use of a "control" group. This control could be arranged such
that each performer in the group performs in the same environment 10 times. This would allow for
correction of the data based on fatigue, and would make the data analysis more conclusive.
Questionnaires
The questionnaires which were given to the singers after each performance set were very carefully
designed to yield a variety of data about the singers' feelings about the environments, in both numerical
and subjective terms. Because no research of this type had been done before, the sheets had to be
designed from scratch, based on expected results. The survey contained a large number of subjective
questions, to allow the musicians to express opinions that were not expected, and thus not numerically
represented on the survey. As the body of knowledge in this field increases, subjective questions should
be de-emphasized, with more attention being paid to numerically analyzable data. In these early stages
of research, the singers' subjective answers serve the purpose of providing revision information for
future questionnaires.
It was obvious in reading the musicians' answers that the questionnaires must be cleverly designed to
pull more information from the singers--to almost "trick" the singers into providing more data. It was
also apparent that the questions should have been phrased much more specifically and precisely. For
example,
46
questions like "how do you think a member of an audience would rate your performance in this setting"
can be made clearer by specifying the type of audience: other singers, for example.
Also, comparing the singers' evaluations with those of the objective listeners was difficult because of the
discrepancies between the two questionnaires. The singer questionnaire focused more on evaluating the
environment, while the listener questionnaire emphasized the singer's performance. Because the focus
of the project should be the performance, not the environment, all references to the effects of the
environment on the singer should be eliminated from the singer's questionnaire. Such questions can
easily be replaced with questions about the performance. Most importantly, all of the numerical
questions on the singer's questionnaire should be on the evaluator questionnaire, and vice versa. All
numerical answers should be based on a percentage scale (0 - 100), with 0 always being the worst case.
Another problem arose because of the three-song structure of the performance sets. The singers were
asked to evaluate themselves after each performance session, which was a more efficient way to manage
the singer's time in the chamber. The objective listeners, on the other hand, rated each song. This
caused resulted in some ambiguity, because in some cases one particular song of a three-song set was
rated lower than the other two. In the future, all evaluations should be done on a session-by-session
basis, or the listeners' individual song evaluations (now on a percentage scale) should be averaged over a
session.
Problems also arose from the inclusion of technical terms, such as "resonant" and "reverberant," in the
singer survey. The singers seemed to latch onto the terms contained in the questionnaire, and used them
questionnaires should contain no technical terms, or the singers should be provided with a glossary of
Based on these observations, a revised singers questionnaire has been created, and can be found in
Appendix 9. This questionnaire should yield a greater quantity of data that is more consistent, and
which can be more easily analyzed. Appendix 10 contains the revised objective listener questionnaire.
ANALYSIS
Time shortages are the enemy of data analysis. Given an infinite amount of time to break down, present,
and analyze data, it would be possible to read a near-infinite amount of information from a finite amount
of data. Given more time, more equipment, and the new questionnaires described in the previous
section, there are several measurements of interest which could be done which were not done in this
report.
The intent of the author, in designing a procedure which involved singing a cappella and with
accompaniment, was to study the effect of environment on tempo. Although the tempo data could be
easily measured with a calculator and a stopwatch, there was not enough time before the due date of this
report to take tempo measurements of all of the performance sets. Therefore the study of the effect of
Another reason for the use of a cappella singing was to measure the effect of environment on pitch.
This analysis was not done because of a lack of equipment, not a lack of time. Frequency analyzers are
rather expensive, and proper pitch measurements would require digitizing the session and measuring the
pitch of specific segments at the beginning, middle, and end of each piece. This is also left to future
research.
48
One other equipment-limited measurement which could not be done in this experiment is the
measurement of the average volume of the singer's voice in each environment. This would require a
49
high-impact microphone at a fixed distance from the singer's mouth, which would be extremely difficult
to keep completely still. Again, this will be left for future researchers.
Finally, something should be said about the method used to analyze and present the numerical data
gathered in this project. The method used to analyze the numerical data in this study is "meta-
statistical." That is, standard statistical analysis of the data presented here would yield a near-zero
significance, and it was decided to create a new method of presentation that would communicate a sense
of possibility of statistical significance, given a larger number of survey participants. Later, larger
studies would have more cause to use traditional statistical analysis, because a larger number of singers
BUDGET
51
The largest frustration in this project was the lack of a budget. If the author had had the foresight and
resources to approach sponsors, to request endowments from the WPI Humanities Department, or to
seek out corporations which would benefit from the results of this research, many of the problems
If the project had had a medium budget ($800 to $5000), the singers could have been paid a stipend, and
would have had more of an incentive to stay through their entire session. More singers could have been
hired. An assistant could have been brought in to help acquire and move equipment. An anechoic
chamber could have been rented for an extended period of time, which would have eliminated the need
to "strike" the equipment after each session. Specific equipment, most notably a signal processing
system designed specifically to simulate rooms, could have been rented, rather than found and borrowed
by happenstance.
If the project were conducted on a large budget ($5000 to $30,000), the author could have conducted the
experiments as at least a part-time job for an extended period of time. A temporary site, independent of
a parent company, could have been set up. The DSP equipment could have been leased for an extended
period. Many more singers could have been hired, resulting in many more data points, and statistically
significant data.
It is the author's recommendation, based on this experience, that future experiments in this area should
be conducted on at least a $1000 budget. A budget proposal for a follow-up project can be found in
Appendix 11.
NEW QUESTIONS
52
This research suggests a number of possible directions for future research in the area of the effects of
acoustics on musical performance. Some of these areas were originally intended to be explored in this
project, before it was realized that the magnitude of the experiment and data analysis was simply too
One parameter which was originally going to be studied in this experiment was frequency response.
Although the room settings did include "High Enhanced" and "High Inhibited" settings, the equipment
used to affect the frequency alteration was not as precise as would have been desirable. Further, an
entire project could be conducted simply to study the effects of the frequency response of a room; this
Another parameter which would merit its own project would be directionality. Although this project
explored the effects of turning off the side speakers and front speakers, there are many, much more
subtle, ways of changing the directional response of a room. Most important among these is the study of
the effects of initial reflections on performance. Not only was the project limited by the DSP equipment
(the only parameter the CP1 has that involves direction is the "Room shape" parameter, which was
shown to have a profound effect on the singers), it was limited to a two-dimensional array of
The idea of a warmup/fatigue curve was touched upon in the performer-by-performer analysis section of
this paper. It is undisputed that singers must warm up their voices. It is also undisputed that, after an
indeterminate period of singing, a performer will become fatigued, then exhausted, and finally hoarse.
Although the last two would be very difficult to study, the transition from "cold" to "warm" to "tired" is
Another interesting question which was raised in the analysis of the objective listeners' surveys, is
whether a singer would evaluate himself on a recording the same way he evaluates himself directly after
a performance. This could be a very valuable phenomenon to study, in determining whether the
psychological effects of acoustics on a performer change his perceptions of the performance. If, for
example, a singer judged a performance as "very bad" directly afterward, but on a recording judged it as
"good," it would be obvious that the singer's opinion was profoundly shifted by the environment.
Related to this phenomenon is the idea of ear coloration. That is, the determination of whether a person
can be forced, over time, to hear a completely unnatural setting as natural--and by extension, whether a
person can be forced to hear a wretched environment as wonderful. Phrased in even more simpler
terms: of a singer same in the same setting five times, would the setting be viewed as progressively
better?
This raises a further question: does warming up involve getting used to an environment, as much as it
Several times, the author has raised the question of the existence of a "peril point:" a point at which the
singer becomes aware of adjusting to a hall's acoustics, and subsequently resets the voice and refuses to
adjust anymore. Time should be spent in determining exactly where this point falls. It is likely that the
point lies in a different place for each performer, and varies with the singer's level of solo experience.
Also, in passing, the question was raised: how would a singer perform if he could not hear his voice at
all? Obviously this is an academic exercise, since bone conduction would always allow the singer to
hear
54
some part of his voice, but the question brings the problems associated with poor concert hall acoustics
Finally, it should be noted that the view of concert hall acoustics presented in this survey may be
erroneous. It is possible that the environment in which a singer performs has no effect on the
performance. Instead, it may be possible that the combination of the present environment and the
previous environment in which a singer performed may have a combined effect. The change from
environment to environment may be more important than the environments themselves. This could be
The above points--the relationship of warm-up to ear coloration, the idea of environmental change
versus absolute environmental conditions--are truly worthy of more study. In the author's opinion, they
are more important than the search for an ideal environment in which to perform. Because if these
phenomena operate as hypothesized, there is no ideal performance environment. Instead, there is a very
specific regimen which a singer can follow, in warming up and in practicing, to make the most of an
environment. And making the most of an environment is the very foundation of the art of performance.
55
56
Works Cited
Backus, John. The Acoustical Foundations of Music. New York: W.W. Norton & Company,
Inc., 1977.
Beranek, Leo L. Acoustics. New York: American Institute of Physics, Inc., 1990.
Bolzinger, S., Warusfel, O., and Kahle, E. "A Study of the influence of room acoustics on piano
performance." Journal de Physique IV 4 May 1994: c5-617.
Campbell, Murray and Greated, Clive. The Musician's Guide to Acoustics. New York: Shirmer
Books, 1988.
Gade, Anders Christian. Musicians' Ideas About Room Acoustical Qualities: An Interview
Survey and a Derivation of Subjective Room Acoustical Parameters. Denmark: The
Acoustics Laboratory, Technical University of Denmark, 1981.
Geerdes, Harold P. Music Facilities: Building, Equipping, and Renovating. Reston, VA: Music
Educators National Conference, 1989.
Jordan, Vilhelm Lassen. "Acoustical Criteria and Acoustical Qualities of Concert Halls." Music
and Room Acoustics. Ed. Royal Institute of Technology. Stockholm: Royal Swedish
Academy. 114-
Lovelace, Austin C. "Good Acoustics for Music and Word." The Hymn 41.3 (July 1990) 15-16.
Olson, Harry F. Music, Physics and Engineering. New York: Dover Publications, Inc., 1967.
Roederer, Juan G. Introduction to the Physics and Psychophysics of Music. Ed. 2. New York:
Springer-Verlag, 1975.
Sundberg, Johan. "Singing and Timbre." Music and Room Acoustics. Ed. Royal Institute of
Technology. Stockholm: Royal Swedish Academy. 57-
57
Appendix 1 - Acoustical Measurements and Standards
Some foreknowledge of acoustical measurement and acoustical standards is assumed in the body of the
text of this paper. To supplement the reader who may not be familiar with certain terms, below is an
appendix of terms associated with acoustical measurement. Much of this information can be found in
Beranek's Acoustics. Those references that can not are referenced individually.
dB or decibel - means that there is a logarithmic ratio between two numbers mentioned, which is
characterized by the term 10*log(N/M). For example, a sound pressure level N, twice as loud as sound
pressure level M, would be said to be 10*log(2/1)=3.01 decibels louder.
SI - Systeme International, or metric, measurement system. We will exclusively use MKS measurement,
meaning all distances will be in Meters, all masses will be in Kilograms, and all times will be in
Seconds.
SPL - Sound Pressure Level. This is a means of determining the 'standard' loudness of a sound. The
unit of Sound Pressure Level is the dbSPL, or decibels of Sound Pressure Level, determined by the
formula SPL = 20*log10 (p/pref), where p is the pressure of the sound being measured, in newtons/m2, and
pref= 2*10-5 newton/m2 for applications dealing with measurements in air, or 0.1 newton/m2 for
calibration of transducers. We will use the first value exclusively in this paper, since we will be dealing
exclusively with air measurements.
T60 - the standardized measure of a hall's reverberance, defined as "the time required for the sound
energy density to decay 60db, that is, to 10-6 of its original value." Note that T60 does not specify what
frequency range should be measured--this is left to the measurer to clarify.
58
Appendix 2 - Letter of Invitation
Dear N,
I am currently working on a six-term, two-part project in audio and acoustics, that may, as a fortunate
byproduct, improve the performability of WPI's Alden Memorial Hall. The first half of the project,
which will be advised by Doug Weeks of the WPI Humanities Department, deals with performers'
perceptions of sound and room acoustics, with a focus on musical and dramatic vocal performance.
I am writing to you because you are a recognized authority in the area of vocal performance, and I
would like to invite you to be a part of the project. Should you choose to become involved, your
subjective observations, combined with that of other performers, would be the basis for the first half of
the project, and would lead to the implementation of an electronic system that will hopefully make
Alden's Great Hall a more hospitable environment in which to perform.
Your role in the project would involve some performance, and analysis of the performances of others, as
they relate to the acoustics of the hall. Because all of the musicians involved undoubtedly have other
time commitments, I have left the project schedule "dangling" until I hear from all of them. The time
commitment will not involve more than two periods (two or three hours each) of performance and
evaluation of the hall, and will likely fall on weekends.
This project gives us an opportunity to improve the lives of musicians at WPI--and if it works out as
planned, musicians around the world. If you are interested in participating, or have any questions about
the project, please contact me. I look forward to hearing from you.
Sincerely,
6 RCA-1/4" adapters
4 BNC-RCA adapters
4 BNC-1/4" adapters
60
Appendix 4 - Chamber Settings
#1 - "Standard" Crown D-75 on fronts, all the way up
EFF LEV -13dB Crown MT2400 on rears, @15
ROOM 4 Crown MT600 on sides, @12
BAL Equal All CP1 gains at max
VOL -8dB 60 degree panorama
AMBIENCE Input level horizontal
SHAPE Rectangle Mic pre-amp gain vertical
LIVENESS 4
ROLLOFF 2.9KHz
PAN 28
POSITION 127
ANGLE 57
#2 - "High Enhanced" Standard, with high EQ all the way up, 14.9KHz rolloff.
#3 - "Most Live" Standard, with LIVENESS = 6.
#4 - "High Inhibited" Standard, with high EQ down, rolloff at 329Hz.
#5 - "Less Presence" EFF LEVEL -20dB
#6 - "Anechoic" SYSTEM/EFFECT MUTE
#7 - "Fan Room" SHAPE Fan
#8 - "No Sides" Sides disconnected
#9 - "Reversed" Mains to rear, Rears to Main
#10 - "Sides Only" Front & Rear disconnected
The room behind that door has been fitted with equipment that makes it possible to adjust its level of
responsiveness from that of a hilltop in the middle of nowhere to the inside of the most reverberant room
in existence, and a variety of different settings in between. It is well-lit and well ventilated, so although
you will be isolated, you will be comfortable.
You will be performing in [x] different environments inside this room. Each performance set will
include two minutes of warm-up in any form you wish. At the end of those two minutes, a tone will
sound. When the tone finishes, please recite the phrase [hand slip] "Joe took father's shoebench out, she
is waiting at my lawn." This is for calibration of the microphones. The tone will sound again. The tone
is tuned to A-440, and should help you fix your own sense of pitch.
While the tone is sounding, please place your guide sheet on the floor of the chamber, where you can
access it at a glance. When the tone ends, please sing the Star Spangled Banner, in whatever key you
wish. Sing it as though you were in a performance situation.
When you have finished singing the Star Spangled Banner, we will give you a minute to catch your
breath and make whatever comments you wish for the recording. We will then start your
accompaniment tape, and you will perform the piece which you chose. Again, sing it as though you
were in a performance situation.
When you have finished singing the piece which you chose, we will give you a minute to catch your
breath and make whatever comments you wish for the recording. We will then start our accompaniment
tape, and you will perform America the Beautiful. Pick the highest octave which is comfortable for you.
Again, sing it as though you were in a performance situation.
When you have finished singing America the Beautiful, we will give you a minute to catch your breath
and make whatever comments you wish for the recording. We will then open the door and bring you out
to fill out a survey based on that room setting. We will then go through this procedure again, starting
with the tone and the Joe Lawn sentence, and ending with the survey outside the chamber.
Do not worry if you miss a note, crack, run out of breath, or have any other sort of vocal glitch. That
data is just as important to the experiment as the performance itself. If you have problems, keep going
to the end of the performance session as well as you can. We will supply water for you to drink between
performance sessions.
If you'd like, I can repeat any or all sections of these instructions for you at any time. If you have any
questions that are not answered by these instructions, write them down and I will answer them after the
entire session is over. If at some point you need a break, feel free to ask and we will take five to ten
63
minutes. However, the reliability of the data in this experiment relies on consistency between
performance sessions.
Here is your first survey sheet. Please complete all questions as fully and clearly as you can. Here is a
cup of room temperature water for you to drink. The amount that you drink now must be the same
amount that you drink during all of the breaks this afternoon, so please remember how much you drink,
and drink enough.
Now we will enter the room, and fit you with a microphone headset. Once it has been fitted, please do
not touch the headset for any reason. The data in the experiment rely on the headset being in the exact
same position throughout each performance. You may remove it between performance sessions.
About twenty seconds after the door closes, the tone will sound, and the performance session will begin.
Make sure you place your guidesheet on the floor, where you can see it at a glance.
64
Appendix 6 - Performance Session Guidesheet
[Reduced to 90% original size]
* ENTER CHAMBER
Acoustically, what would you say this setting lacked? How did you notice it?
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
_______________________________________________________________________________
Acoustically, what would you say this setting did not lack? How did you notice it?
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
_______________________________________________________________________________
How do you think you adjusted, in order to sound your best in this environment?
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
_______________________________________________________________________________
If you could change one specific acoustical property of this environment to make it better, what would it
be? Why did you choose that particular property?
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
_______________________________________________________________________________
Approximately how many times have you sung the Star Spangled Banner? _____
Approximately how many times have you sung America the Beautiful? _____
Approximately how many times have you sung your song of choice? _____
66
Does this setting remind you of any halls you have sung in, or listened to music in, before?
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
_______________________________________________________________________________
How do you think a member of an audience would rate your performance in this setting?
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
_______________________________________________________________________________
How would you critique your performance in this setting? What mistakes did you catch yourself
making [including momentary vocal glitches]?
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
_______________________________________________________________________________
67
Appendix 8 - Performance Evaluation Sheet
While listening to the three pieces, please tabulate the following items, and note unusual features of each
performance in the space provided. If you need more space, please use the back of the sheets, and be
sure to indicate what number/letter you are extending.
What vocal adjustments/performance characteristics make this performance different from all or some of
the other performances? Any other comments?
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
________________________________________________________________________________
68
What vocal adjustments/performance characteristics make this performance different from all or some of
the other performances? Any other comments?
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
________________________________________________________________________________
69
What vocal adjustments/performance characteristics make this performance different from all or some of
the other performances? Any other comments?
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
_______________________________________________________________________________
70
Appendix 9 - Revised Questionnaire
Please complete the following survey, which contains questions about the performance you just
completed. Please be as thorough and as precise as you can. If you need more space, please use
the back of this sheet, and be sure to indicate which question you are extending.
How would you change the acoustics of this room, to make it more friendly?
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
_______________________________________________________________________________
How did you adjust in order to sound your best in this room, if at all?
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
_______________________________________________________________________________
How would you describe this room physically (size, materials, architectural features)?
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
_______________________________________________________________________________
71
Please answer the following questions using a percentage scale, where 0% is the worst case, and
100% is the best.
Please rate your performance in terms of the following criteria. Use a percentage scale, where 0%
is an absolute failure and 100% is a flawless performance.
How would you rate this room, if it were a concert hall? _____%
How do you think another musician would rate your performance, overall? _____%
How do you think a member of an audience of singers would rate your performance?
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
____________________________________________________________________________________
_______________________________________________________________________________
72
Appendix 10 - Revised Performance Evaluation Sheet
While listening to the performance sets (1-10), please evaluate the performances in terms of the
following criteria. Use a percentage scale, where 0% is an absolute failure, and 100% is a flawless
performance. Use the extra blank space provided to note unusual features of each performance.
OR
TOTAL $3215
TOTAL $21,190