Вы находитесь на странице: 1из 11

Stereophony

a series of perspectives

Richard Hallum, M. Mus. Tech (merit)1


The Stereo Listening Angle

An interesting assumption with stereo is that the playback geometry is an
equilateral triangle between the two loudspeakers and the listener. Martin [1]
specifies this configuration in his paper on stereo microphone techniques, but
many authors make no mention of it. Moore [2] spends thirteen pages describing
the psychoacoustics of Hi-Fi sound reproduction, covering frequency response
and phase response in some detail but spatial aspects are completely ignored
(even though a whole chapter is dedicated to space perception).

Don't use any other angle other than 30 degrees for stereophonic monitoring. The
laws of physics determined this for stereophonic listening 50 years ago.

So wrote Carl Tatz [3] in Tape Op magazine. Any checking of textbooks or the
Internet will quickly confirm that the idea of an equilateral triangle between the
two loudspeakers and the listener has long since been accepted as the best
geometry for stereo playback: eg Stereo playback assumes a symmetrical
loudspeaker and listener setup with 600 degree listening angle between the
loudspeakers and corresponding to an equilateral triangle configuration.
Linkwitz [4]. The reason for this seems to have been lost in the mist of time, so I
would like to revisit this matter with two questions:
Why was 60 chosen, and, is 60 the optimum angle?

It is a reasonable assumption to think that the 60 angle was set in the early days
of stereo. After all, stereo was not a matter of companies simply wanting to add
an extra loudspeaker as a sales gimmick. Rather, the whole idea of stereo was
considered theoretically and experimentally in the 1930s. There is evidence to
suggest, however, that locking the listening position at 60 was a much later
consideration.
My own first experience with stereophonic reproduction was from a La Gloria
stereogram in the mid 1960s. This was a very typical unit for the time which
conveniently placed all the components (record deck, tape deck, tuner/amplifier,
and speakers) in a single cabinet. This meant that the integrated loudspeakers
were placed at either end with a distance between centres of about 90cm. The
drivers were mounted in-line with the front of the cabinet, so their axes were at
90. Admittedly, cosmetics and/or convenience have often overshadowed
technical considerations for domestic loudspeakers so this example doesnt
really prove that the 60 angle was not a consideration at or before that time.
Lets take another example: The JBL L100 loudspeaker is the consumer version
of the very popular 4310 studio monitor (more than 125,000 pairs were sold).
By the end of the 1970s, recording studios in the United States used more JBL

1 Senior Tutor, Music & Audio Institute of New Zealand
monitors than all other brands combined. It is surprising then to find these
comments in the 1970 L100 manual [5]:

For the best possible stereo performance, the two loudspeaker systems should be
arranged symmetrically on each side of the listener. As a general rule, a person
sitting in the general listening position should see an angle of about 40 between
the two sound sources. The distance from one loudspeaker enclosure to the other is
determined by their distance from the listener and by the 40 listening angle.
A drawing clearly showing a 40 angle between the loudspeakers is shown, and
the caption text states: Sound energy from each loudspeaker blends to form a
stereo wall of sound 2. The stereo image will be intensified and the area of best
stereo perception increased if the two systems are rotated slightly toward the
preferred listening position.
Clearly, JBL are stating that an isosceles triangle with a 40 angle is best.
The last sentence is rather telling, in that it accepts that the listening distance
may change and therefore the loudspeaker angle may be adjusted so that they
are not on-axis to the listener. This is considered in more detail later.

The origins of stereo did not dictate a 60 angle, and that for the domestic
market, at least, other factors of loudspeaker placement were and are important
(often to the detriment of the stereo image).
JBL have always been an acknowledged leader in professional loudspeakers for
livesound and studio applications, but the L100 was a consumer product, so now
let us consider professional installations: studio monitors in designed control
rooms.
Several standards bodies have published recommendations for monitor setup,
namely AES20 (1996, revised 2007), EBU3276 (1998), ITU BS1116-1 (1994,
revised 1997). These all share a relationship with IEC 60268-13, which applies to
domestic setups.3

Table I
Organisation/Standard Listening Separation Wall Height Listening area
distance^ angle distance (from floor)
AES20 > 1m Not < 40 > 1m Not Not specified
< 3m Ideally 60 specified
EBU3276 > 2m Optimum 60 > 1m > 1.2m Radius
< 4m < 0.8m
ITU BS1116-1 > 2m+ 60 > 1m > 1.2m Not specified
< 3m*
^ listening distance is to the loudspeaker baffle (not the median plane distance)
* up to 4m allowable in suitably designed rooms
+ limits of listening distance are D = 2 to 1.7 B (m), where B is the baseline (distance between the
speakers)


2 a JBL marketing term (nothing to do with Phil Spector, or The Grateful Dead).
3 IEC60268-13 specifies an aspect ratio of (w/h) (l/h) (4.5(w/h -4)), l/h < 3 w/h < 3


The angle at the extremities of the ranges allowed by these standards is given in
the following table:

Table II
Listening Separation Loudspeaker
distance angle separation
AES20 1m 40 0.68m
1m 60 1m
3m 40 2.06m
3m 60 3m
EBU3276 2m 60 2m
4m 60 4m
ITU BS1116-1 2m 60 2m
3m 60 3m
4m 60 4m

Christensen [6], states that the Listening position and loudspeakers should form a 60
degrees triangle. This is an old practical rule. Wider distance between speakers
makes it difficult to have a stable phantom image center. Smaller distance between
speakers gives an incorrect stereo image, with too little space between the players.

Bech [7] experimented with listener preference of stereo width, using 30,
22.5, and 12. He concludes that:

1. The quality of reproduction of space for both audio-visual and audio-only


program material increases with increasing stereophonic width. The influence
was found to be independent of the audio-video program and the rate of
change was found to be largest for the audio-only programme.

2. The reproduction level has a significant influence on the quality of reproduction of


space. The quality of space increases with increasing reproduction level
independent of stereophonic width and subject position.

3. The listening position has a significant influence in the quality of reproduction of


space for audio-visual material. The quality of space decreases as the
listening position moves off-centre independent of reproduction level and
stereophonic width.

4. The extra low frequency energy added by a sub-woofer does not have a significant
effect on the quality of reproduction of space.


However, Gerhard et al [8] suggest that a wider angle is optimum: The resulting
subtended angle should be in the 70 to 90 range. In a companion paper [9] it is
shown that the centre image (considered to be the weakest part of a two
loudspeaker stereo sound stage) in a 74/low reflection-layout can be more stable
than the two thirds off-centre image of a 60/high-reflection layout.

The possibility of angles wider than 60 was considered at the onset of stereo
playback. In a 1934 paper, Steinberg and Snow [10] explained that a centre
loudspeaker would prevent the hole in the middle problem:

The three-channel system proved definitely superior to the two-channel by


eliminating the recession of the center-stage positions and in reducing the
differences in localization for various observing positions.

Audyssey Laboratories [11] have come up with an interesting front speaker


solution, as part of their DSX surround system which has 11.2 channels. The left
and right front loudspeakers at 45 and left and right wide loudspeakers at
60.


fig. 1

Toe-out.
An interesting discrepancy now exists between studio monitor and domestic hifi
loudspeaker geometry. Almost all studio monitors are positioned on-axis to the
listening position, whereas home hifi loudspeakers are often placed parallel to
the room axis. In the more casual listening environment of the home it could be
assumed that the single reason for this is the tidier look achieved (as most
furniture is positioned in the room parallel to the walls). There is, however, a
sound consideration for this setup.
In order to get a realistic soundstage the distance between the loudspeakers and
the listener must be reasonably large. Close distance will mean a person is
listening to the individual loudspeakers, and the sense of a good stereo image
will not occur. If an equilateral triangle layout is preserved, then in a typically
sized living room, the loudspeakers will be quite close to the sidewalls. This
results in the power response of the loudspeaker being more important than the
on-axis frequency response. A toe-out of 30 will mean the reflected sound is less
off-axis as it leaves the loudspeaker. It will also increase the ratio of reflected to
direct sound. This is not necessarily a problem as lateral early reflections are
able to enhance the stereo image, however Linkwitz [12] suggests that room
reflections effect depth perception of phantom images. Note that this is not the
case in a critical monitoring situation, where lateral reflections are carefully
controlled (being delayed or eliminated).
Given that hifi users are often providing toe-out to their setups, it follows that
some loudspeaker manufactures are designing with that in mind. Modern
crossover and baffle diffraction modelling have enabled hifi loudspeaker
manufacturers to design the acoustic axis to differ from the traditional 90 off
the baffle.
On-axis frequency response may be too bright (depending on baffle step
compensation), or loudspeakers may actually present their flattest frequency
response when off-axis.
Few loudspeaker manufacturers are willing to publish meaningful data on
dispersion (directivity) so users must resort to trial and error to ensure the
optimum choices are made in loudspeaker selection and positioning. Geddes [13]
states that this area of loudspeaker design has been grossly overlooked in the
vast body of literature on loudspeaker design. It is also the case that at some
angle (preferably on-axis) the loudspeakers must have an even and equal
frequency response, or accurate rendering of a stereo image will not be possible.
Linkwitz & Barrington [14] touch on this matter in paper on Dipoles.

Although localisation was originally the top focus in stereo (or 3 channel)
systems, comparatively little recent research has focused on it. Mason [15]
reviewed over 40 papers and found that subjective attributes other than
localisation accounted for 71% of elicited responses. An interesting side-point is
that without a substantial body of information on stereo localisation one
wonders whether this will impede 3D multi-channel localisation design and
development.
Foo et al [16] researched localisation using a variety of loudspeaker angles (from
20 to 110) including asymmetric setups. The signal source was spoken word
encoded with HRTFs to produce signals that emanated from twelve points
equidistant around the listener. Of relevance is that the subjective judgement of
sound direction was consistently good between +90 and -90 for both the
subtended 40 and 80 angles.

Generally, listeners would consider stereo to sound good if it provides a
sufficient sense of spaciousness. In most cases, providing any form a stereo over
mono will provide a pleasant enough experience for the listener. In this case an
arbitrary 60 playback angle will suffice. However, for stereo to be as accurate as
possible it should maintain the correct sense of position (in terms of angle and
depth) for each sound, as it was performed. Thus, we have a conundrum: an
absolute recording angle is required to set the correct playback angle, and an
absolute playback angle is required to gauge the right recording angle. This is a
companion problem to Floyd Tooles Circle of Confusion [17].

2.0 0.7

44
30

44

4.0





fig. 2



Table III
Angle () Spacing (m) Distance (m) ratio
45 3.0 3.92 1.31
60 3.0 3.0 1
90 3.0 2.12 0.707
120 3.0 1.73 0.577
Changes in distance with angle


70.00

60.00

50.00
Apparent Angle ()

40.00 45

True Angle
30.00
60

20.00

10.00

0.00
0 5 10 15 20 25 30 35 40 45 50


fig. 3 Angular error for 45 and 60 loudspeaker angles



Summary
It has been found that the 60 playback angle was not described in original
stereo proposals. It eventually became a de facto standard. Beyond 60 the
stereo width soon becomes unnaturally wide, and will exhibit a hole in the
middle. It also becomes impossible to achieve (at a sensible listening distance)
in a typical living room.
Narrower angles put the listening position too far back. This has the effect of
increasing reflected sound which blurs the already reduced stereo image.
Using a 60 in a typical concert hall puts the listener in a seating position that
will be at least halfway back in the stalls.
Although it is a compromise, the benefits outweigh the disadvantages, namely:
1. It provides a reference point for stereo recording techniques
2. It provides a stereo reference for studio mixing
3. It provides a simple setup for domestic users who wish to experience the
stereo recording as intended.
Points 1 and 2 are essential for recording professionals. For amateur listeners, it
is simply a matter of taste. In the same way that one might select close seats at a
concert if a wide soundstage is preferred, they are free to place loudspeakers as
desired.

References:

[1] Martin, G. (2003). General Response Characteristics of Microphone Configurations, AES
114th Conv.

[2] Moore, B. (2003). An Introduction to the Psychology of Hearing, 5th ed., Academic Press.

[3] Tatz C. (2010). Monitor Position, Tape Op Magazine, Issue #78.

[4] Linkwitz S. (n.d.). Theory Mapping from recording to playback
http://www.linkwitzlab.com/Recording/record-play-map.htm

[5] JBL. (1970). JBL L100 Loudspeaker Manual
https://www.hifiengine.com/manual_library/jbl/l100.shtml

[6] Christensen O. L. (1996). A Practical Guide to Acoustical Design of Control Rooms and
Placement of Loudspeakers. AES 100th Conv.

[7] Bech S. (1997). The Influence of Stereophonic Width on the Perceived Quality of an Audio-
Visual Presentation Using a Multichannel Sound System. AES 102nd Conv.

[8] Gerhard et al. (1996). Loudspeaker Placement for Optimised Phantom Source
Reproduction. AES 100th Conv.

[9] TheiB B. & Hawksford M. (1996). Localization Experiments in Three-Dimensional Sound
Reproduction. AES 100th Conv.

[10] Steinberg J. & Snow W. (1934). Audio Perspective Physical Factors. Electrical
Engineering, Jan. 1934

[11] Silva R. (2017). The Audyssey DSX Surround Sound Format.


https://www.lifewire.com/audyssey-dsx-surround-sound-format-1846862

[12] Linkwitz S. (2009). The Challenge to Find the Optimum Radiation Pattern and Placement
of Stereo Loudspeakers in a Room for the Creation of Phantom Sources and
Simultaneous Masking of Real Sources. AES 126th Conv.

[13] Geddes E. Audio Transducers (eBook).


http://www.gedlee.com/Books/AudioTransducers.aspx

[14] Linkwitz S. & Barringer D. (2009). Recording and Reproduction over Two Loudspeakers
as Heard Live: Part 1: Hearing, Loudspeakers and Rooms. AES 126th Conv.

[15] Mason R. (2017) How important is accurate localisation in reproduced sound? AES 142nd
Conv.

[16] Foo K. et al. (1999). Optimization of Virtual Sound Reproduced using Two Loudspeakers.
AES 16th Conference.

[17] Toole F. E. (2008) Sound Reproduction Loudspeakers and Rooms. Focal Press


Appendix

Intensity Stereo

18

!1
30

!a

"#$%& )*+
Bauers Stereo Law of Sines states that =
"#$%' ),+


1 = 6
+
For L=1, R=1, sin1 = 0 so 1 = 0

For L=1, R=0.5, sin1 = 0.166 so 1 = 9.6 L


Intensity difference for two coincident microphones is given by the formula


+ Q 1
= 201M 2
Q
+ + 1
2

where m is the angle between the microphones.

Assuming a pair of Cardioid microphones angled at 90, then reproduction from
two loudspeakers will give:

and


L' R'
L R



Sound to the left ear will be
+ h = + ( )

where T is the delay time for the R signal.

= + ()

S

DT
!

0.2

Taking the effective head-width to be 0.2m, then delay time to the right ear will
be
q 0.2
= = (mS) = 0.5831sin
0.343

The corresponding phase difference will be

= 0.58 10*y

The phase angle at the listener will be

z z
=
2 + z + z

At low frequencies, will be small, so that

)*+
= T
),+

and
+
=
+ + +

therefore
= ()

and
= 0.58 10*y 6

Substituting for and T gives:

= | 6

which is the error between the true and apparent angle.

Вам также может понравиться