aaaaCzęsto usiłujemy ukryć nasze uczucia przed tymi, którzy powinni je poznać.aaaa [ Pobierz całość w formacie PDF ]
Applied Acoustics 62 (2001) 91±108
www.elsevier.com/locate/apacoust
Spatial responsiveness in concert halls and the
origins of spatial impression
A.H. Marshall
a
, M. Barron
b,
*
a
b
Department of Architecture and Civil Engineering, University of Bath, Bath, Somerset, BA2 7AY, UK
Received 16 September 1999; received in revised form 6 March 2000; accepted 27 June 2000
Abstract
The story of research into spatial eects in auditoria is an intriguing one. Serious studies
only began with the development in the 1950s of simulation systems which reproduced direct
sound, early re¯ections and reverberation in anechoic chambers. The traditional view had
been that spatial eects were associated with later reverberation. This paper discusses the
early work from the late `60s on spatial eects produced by early lateral re¯ections from the
perspective of two early participants. Subsequent work on spatial impression and the impor-
tance it has for overall concert hall acoustics are also considered. The current position is that
two or more spatial eects may coexist in concert hall listening and that the nature of the
performance, from solo to fully scored orchestra, may in¯uence the spatial eects experienced.
#
2000 Elsevier Science Ltd. All rights reserved.
Keywords: Concert hall acoustics; Spatial hearing; Spatial responsiveness; Spatial impression; Source
broadening; Listener envelopment
1. Introduction
Spatial hearing is an elusive sense. It is of course dominated by localisation, so
that with a quiet sound only the direction of the source is perceived [1]. With louder
sounds, we also become aware of re¯ected sound arriving from other directions and
can use this information to make judgements about room size and acoustic char-
acter. This sense of room sound is usually unconscious whenever the sound matches
* Corresponding author. Tel.: +44-1225-826-826; fax: +44-1225-826-691.
E-mail address: m.barron@bath.ac.uk (M. Barron).
0003-682X/01/$ - see front matter
#
2000 Elsevier Science Ltd. All rights reserved.
PII: S0003-682X(00)00050-5
Tuatapere R.D., New Zealand
92
A.H. Marshall, M. Barron / Applied Acoustics 62 (2001) 91±108
expectation. With variable intensity sounds like music, the elusiveness of spatial
eects arises partially because the spatial character varies continuously with level.
Spatial sound was originally associated with reverberation and was studied in some
interesting experiments during the 1960s [2]. Then in the late `60s it was proposed that
there was an additional spatial phenomenon associated with early re¯ections from
the side, frequently called spatial impression. Since that date the majority of studies
of spatial hearing in rooms have concentrated on this spatial impression associated
with early re¯ections. Though the spatial eect of reverberation was not denied, it
has remained rather a mystery. In the last 5 years, it has been suggested that there are
two spatial eects, that of source broadening due mainly to early re¯ections, and that
of envelopment linked principally to later reverberant sound. But with two separate
spatial eects both ¯uctuating with sound level, mysteries are perhaps predictable.
2. Early experiments with auditorium simulation systems
During the ®rst half of the 20th century, it was gradually realised that acoustic
quality in auditoria both for speech and music depended on more than reverberation
time alone. While qualitative comments were made, no new objective measures from
before 1950 have survived. Comments were made that the early re¯ections were
signi®cant but it was only with the development of simulation systems that the role
of early re¯ections began to be quanti®ed. These simulation systems involved loud-
speakers mounted around the listener in an anechoic chamber. An anechoically
recorded signal is fed to a speaker in front of the listener to reproduce the direct
sound; for re¯ections, delays were generated with tape loops and reverberation
chambers or plates were used to simulate reverberation.
Subjective testing using a simulation system was pioneered at the University of
GoÈ ttingen under the direction of Professor Erwin Meyer. One of their ®rst publications
by Haas in 1951 [3] has been the key to the design of public address systems ever
since. Much of the GoÈ ttingen work during the 1950s and early 1960s was summarised
in English in reference [4]. With hindsight one can see that many of their experi-
ments were directed at solving one basic problem: how far can the complex impulse
responses found in rooms be simpli®ed without changing the audible impression?
Two approaches to interpretation of impulse responses were pursued simulta-
neously: the statistical and energy approach. The statistical approach concentrates
on the relative levels and delays of re¯ections. Early analyses involved categorising
re¯ection sequences in terms of the number of re¯ections with particular levels (such
as more than
ÿ
6 dB) relative to the direct sound. Schodder [5] introduced this ana-
lysis in 1956 and Junius [6] 2 years later described an apparatus for automatic sta-
tistical analysis (no mean feat with only valves/vacuum tubes at his disposal). From
Russia, Preizer [7] undertook similar studies in 1966.
The energy approach is based on the principle that subjective response is related to
measures involving integrated re¯ection energy. Thiele [8] in 1953 suggested the 50
ms early energy fraction as a measure of clarity or distinctness, `Deutlichkeit' in
German:
A.H. Marshall, M. Barron / Applied Acoustics 62 (2001) 91±108
93
50
ms
1
0
p
2
t
:d
t
Deutlichkeit;
D
p
2
t
:d
t
=
1
0
The correlation of D with speech intelligibility was established in 1956 by Bore [9,
p.190] and the quantity is still used today. For music, Beranek and Schultz in
America in 1965 [10] were also suggesting that this energy ratio was signi®cant. The
logarithmic ratio of early-to-late energy with the early sound integrated over 80 ms,
frequently referred to as the Clarity Index, C
80
, is now the recommended measure
for music (®rst explicitly proposed by Reichardt et al. [11]).
A threshold measurement oers a subjective test with few prior assumptions.
Threshold measurements were common in the early days of simulation experiments
and those by Burgtorf [12] and Seraphim [13] from 1961, both working in GoÈ ttingen,
are particularly important for the story of spatial impression. Seraphim conducted
threshold experiments on re¯ections with speech as the signal. Again one aim of this
work was to discover how much detail in the impulse response was ignored by listeners.
The thresholds (such as Fig. 1) showed that the relative direction and delay of
re¯ections was signi®cant. This result implies the necessity of applying a statistical
analysis to the impulse response.
When re¯ection threshold measurements with music were actually undertaken by
Schubert in 1966 [14], it was found that for continuous legato music the threshold of
a re¯ection was much less sensitive to delay. The relative arrival time of a re¯ection
is thus much more important for speech than it is for most types of music.
Fig. 1. Thresholds of re¯ections with speech for various directions (after Seraphim [13]). O, primary; I,
masking; T, masked re¯ection.
94
A.H. Marshall, M. Barron / Applied Acoustics 62 (2001) 91±108
Returning to the competing approaches, the statistical versus the energy
approach, the statistical approach with one exception is seldom discussed today. The
exception is disturbance by echoes, already in 1952 [15] it was noted that disturbance
by a late echo is reduced if an earlier re¯ection is added. Apart from echo dis-
turbance, the energy approach is involved in all current well-known measures for
music auditoria [16], but in time this might be seen as an oversimpli®cation.
3. The early history of spatial impression
The earliest casual reference to the eects of early lateral re¯ections of which we
know was published in 1952 by Meyer and Schodder [17]: ``
...
the presence of a
secondary loudspeaker creates an apparent enlargement of the spatial extent of the
primary source and with a delay of some 10 ms also a certain `reverberance' ''
(translation from German MB). Neither the authors nor their co-workers in GoÈ t-
tingen appear to have realised the signi®cance of this at the time. For instance, from
the experimental work at GoÈ ttingen, Lothar Cremer had appreciated the importance
of early re¯ections in concert halls and had ensured their presence in the design of
the Berlin Philharmonie of 1963. However, he was not aware of the signi®cance of
early re¯ection direction, though there are of course several locations in the Phil-
harmonie where surfaces surrounding the audience supply lateral re¯ections [18, p.
256]. (It is also interesting to observe in the quote above that the spatial eect pro-
duced by the lateral re¯ection is linked by the authors to spatial aspects of rever-
beration, a subjective overlap which is still not fully understood today.)
The major publication on concert hall acoustics of the 1960s was Beranek's
``Music, acoustics and architecture'' of 1962 [19]. Beranek had conducted a survey of
54 concert halls and opera houses, which are described in the book with scale
drawings, photographs and available acoustic data. For subjective assessments of
the halls, Beranek held interviews with leading conductors and performers. These
data were used to allocate the halls into ®ve categories in terms of their acoustic
quality. Beranek then sought to correlate the subjective ratings with objective
acoustic or physical data for the individual halls. Whereas historically most atten-
tion had been paid to the reverberation time of halls, Beranek found that the delay
of the ®rst re¯ection, the initial-time-delay gap, was the most important correlate of
quality. He considered the preferred delay to be less than 20 ms. Beranek suggested
that initial-time-delay gap was linked to perceived intimacy, the sense of whether the
performance is taking place in a small space or a large one; an intimate hall can also
be described as having `presence'. In Beranek's rating scheme, the initial-time-delay
gap contributed to 40% of the total rating; by contrast for reverberation time the
proportion of the total rating was only 15%.
The publication of Beranek's book coincided with the opening of the ill-fated
Philharmonic Hall in New York. Indeed Beranek's survey had been accelerated to
provide data to inform the design of Philharmonic Hall. The design as built (which
Beranek subsequently repudiated as deviating substantially from his recommenda-
tion) sought to achieve the desired short initial-time-delay gap by suspending an
A.H. Marshall, M. Barron / Applied Acoustics 62 (2001) 91±108
95
array of re¯ecting panels well below the high ceiling and over about half the plan
area. With hindsight one can see that this design missed the point that the direction
of these re¯ections was important.
The management of the Philharmonic Hall subsequently engaged several acousticians
to advise on the reasons for the disappointing acoustics in the hall. One by-product of
this work is worth recording here: measurements in the Philharmonic Hall were
responsible for the discovery of the attenuation at grazing incidence when sound
passes over seating, also known as the seat-dip eect [20]. The results of more gen-
eral acoustical measurements in the hall by Schroeder, Atal, Sessler and West [21]
from Bell Telephone Laboratories were published in 1966. Included in the paper is
an objective measure, surprising for the time, related to the ``directional distribution
factor of the early sound energy''. They measured the proportion of early sound
coming from the side and tentatively concluded ``that the directional distribution of
early re¯ections is a signi®cant contributing factor to acoustical quality''.
Marshall's ®rst publication on the possible importance of early lateral re¯ections
had its genesis due to his involvement in the design of a new hall for the city of
Christchurch in southern New Zealand. The story is elaborated in the next section
below; his paper [22] was submitted for publication in August 1966. Marshall was sur-
prised by the paucity of advice available concerning the fundamental architectural
question of preferred room shape. As well as exploring this question in the paper,
the quality of `spatial responsiveness' was proposed which Marshall suggested is
present when the lateral re¯ections are audible and not masked. Using the only
available threshold data at the time by Seraphim for speech [13], the need for
unmasked lateral re¯ections led to the proposed requirement for lateral re¯ections to
arrive before the overhead re¯ections in order for the spatial eect to be perceived.
In November 1966 West gave a paper [23] at the 72nd meeting of the Acoustical
Society of America. With an abstract so brief, it is easiest to quote it in full.
For positions along the center aisle of a rectangular hall, the ratio of travel times of
early re¯ections from the ceiling and walls is approximately equal to 2 H/W,where
H is the height and W is the width of the hall. This ratio was determined for 38
concert halls and opera houses and the results were compared with the subjective
categories compiled in Beranek's extensive investigation [19]. The correlation coef-
®cient for 2 H/W and a numerical scale of the subjective categories is 0.71. Thus
it appears that this ratio, which determines the relative arrival times of early lateral
and vertical re¯ections, accounts for a major portion of the subjective ratings. The
importance of lateral versus vertical re¯ections has been substantiated by labora-
tory experiments conducted in an anechoic chamber, using delayed speech and
music signals played through loudspeakers.
In other words, both Marshall and West independently were implying that Bera-
nek's highly rated halls had ®rst re¯ections which both arrived early and from the
side, not from above. In the next year 1967, Marshall submitted a note [24]
responding to new threshold data published by Schubert [14] for re¯ections with
music. The importance of bass frequencies for the spatial eect of early lateral
re¯ections was highlighted for the ®rst time here by Marshall, as was the basic
[ Pobierz całość w formacie PDF ]