March 29, 2014

Investigation into Tolerance of Home-theaters’ Surround Speaker Placement for Acceptable Perception of Reproduced Sound



As WG Group under Organization of JAS(Japan Audio Society)

Mick M.Sawaguchi
Shinji Koyano
Kazuhiko Terui
Satoshi Suzuki
Atsushi Marui
Bike H.Suzuki




Surround sound is normally produced with the loudspeakers arranged according to the recommendation ITU-R BS.775-2. However, in ordinary homes, especially in Japan, the produced surround sound is not necessarily enjoyed with the same loudspeaker arrangement. Investigation and research was performed to study the tolerance of surround speaker arrangement for listeners at home to enjoy the surround sound with the quality level as close as possible to what the original production intended for.
The goal was to develop and propose a practical guideline for surround loudspeaker placement in designing a good home-theater.




1. INTRODUCTION
It is normally recommended that surround loudspeakers of a home-theater audio system should be placed according to the ITU-R Recommendation BS.775-2, Multichannel stereophonic sound system with and without accompanying picture.  (Fig.1)The recommendation is based on the agreement on the results of evaluation tests of surround loudspeaker placements performed globally in the 1990’s. Most rooms specifically for surround sound production have the surround loudspeakers placed right according to the ITU-R recommendation. 

If the same loudspeaker placement can be installed at ordinary homes, users would be able to enjoy the sound as it was intended in the production field. 
However, in Japan rooms at homes are forced to have several restrictions in terms of room size and shape. Many home-theater owners must be content with a loudspeaker placement which is very different from the one recommended by ITU.
The research presented here was performed to clarify the torelance of surround loudspeaker placement for Acceptable perception of reproduced sound.

The research consisted of investigation of actual loudspeaker placement at homes, and auditory impression evaluation test, which made comparisons of the auditory impression between the ITU-R Recommendation and actual domestic placements. 



2. SURVEYS ON LOUDSPEAKER PLACEMENT IN HOME-THEATERS

2.1 Gathering and Analysis of Samples of Loudspeaker Placement

Home theater installers and individual home-theater owners offered a total of 82 samples of data on loudspeaker placement in home theaters. The samples of data included:

1. Locations of surround loudspeakers on a plane: distance from the listening position in meters, relative position to the listening position, and position of display or screen.

2. Height of each loudspeaker (from floor to the physical center of the loudspeaker)

3. Sketch drawings or pictures of loudspeaker placement

From the data above the following parameters were calculated:

1. The plane-projected distance and the real distance from the listening position at a height of 1.2m, for each loudspeaker except for a subwoofer.

2. Plane-projected angles between loudspeakers:
Left Ch. (L) to Center Ch. (C)
Right Ch. (R) to Center Ch. (C)
Left Surround Ch. (LS) to Center Ch. (C)
Right Surround Ch. (RS) to Center Ch. (C)

3. Slant angle (upward or downward) from the listening position, for each loudspeaker.

2.2 Typical Loudspeaker Placement Patterns Extracted

Of 82 samples, 69 samples which deal with 5 surround loudspeakers were subjected to cluster analysis with three parameters: (1) plane-projected distance of each loudspeaker, (2) plane-projected angles between loudspeakers, and (3) slant angles (upward or downward) from the listening position.
As a result, the 69 samples of loudspeaker placement were grouped into the following five patterns:

Pattern 1: All loudspeakers are positioned higher than the listening position.

Pattern 2:  Only rear loudspeakers are positioned higher than the listening position.

Pattern 3: Front loudspeakers are narrower angled, rear wider angled.
Pattern 4: Similar to the placement by ITU-R recommendation
Pattern 5: Rear loudspeakers are positioned right at the sides of the listener





                                                    Fig.2 Five Typical Patterns of Loudspeaker Placement

3. Perceptual Audio Evaluation Test

3.1 Selection of loudspeaker placement patterns for evaluation test


Sample recording room under IEC Spec



Two more patterns were added to the five (5) patterns for the auditory impression evaluation test. The additional two patterns allow for the growing trend in Japan to place loudspeakers in the ceiling of home theaters.

The seven (7) patterns are:

Pattern 1: Rear loudspeakers are angled excessively wide from the listening position
Pattern 2: Rear loudspeakers are positioned higher than the listening point. It is assumed that many home theaters have rear loudspeakers installed on the walls.
Pattern 3: The front center loudspeaker is positioned lower than Left, Right loudspeakers and the listening point. The pattern corresponds to a case where front center loudspeaker is positioned below a large screen.
Pattern 4: Rear loudspeakers are positioned right at the sides of the listener, and the listening point is very close to the rear wall.
Pattern 5: The three front loudspeakers are positioned lower than the listening point. The pattern accounts for home theater products called “Theater Rack.”
Pattern6: Assumed is a case where the rear loudspeakers are installed in the ceiling
Pattern 7: Assumed is a case where all the loudspeakers are installed in the ceiling.

                         Fig.3 Loudspeaker Placement Patterns




Evaluation test sample recording with 7 deferent SP patterns by Dummy Head



3.2 Evaluation Attributes and Sound Samples

From the viewpoint of evaluating expressly and clearly fields of surround sound, three (3) evaluation attributes were selected: (1) Listener Envelopment (Sensation of being embraced by sounds), (2) Sound Image Localization, and (3) Sensation of movement.
For each evaluation attribute, the following three sound samples were prepared:

(1) Sounds recorded in a natural outdoor environment (for Listener Environment)
     Early in the morning, in a wood. Birds are singing all around over 360 degrees.
(2) Instrumental Music (for Sound Image Localization)
     Main Vocal in the center. Instruments positioned all over around 360 degrees.
(3) Sound Effect  (for Sensation of movement)
     A single sound source rotates around the listener in increasing speeds on a horizontal plain.

Editing was made so that the duration of each sound sample is one minute.

3.3 Preparation of sound samples for evaluation

It is impractical to set up all the seven loudspeaker placement patterns for each evaluation because of the time and labor encountered: subjects must wait for a certain time before a new pattern of loudspeaker placement is set up, and tend to forget the impression of the sound presented in the preceding pattern.
In order to reduce the load on subjects and implement higher evaluation precision by presenting sound samples freely and instantly, the evaluation test was designed to use a method of presenting the sound samples, which are recorded by a dummy head. In preparation for this, binaural recordings of the sound samples described in 3.2 were made in a reference listening room, which is in conformity with the ITU-R BS.1161-1, through a dummy head to produce sound samples for the evaluation test.                                                          

3.4 Evaluation Method

Subjects of the evaluation were 25 adults in their twenties to fifties who are audio specialists with aurally normal ears. The 5-point rating scale method was used in the evaluation test: Subjects evaluate each sound sample for (1) listener environment (sensation of being embraced by sounds), (2) sound image localization, and (3) sensation of movement, by marking a level in five levels with “3” as the level which means that the evaluated quality of a sound sample is equal to that of the reference sound sample. Since the ratings were made in comparison to the reference stimulus, which is set to 3.0 on the rating scales, an average rating beyond 3.0 for a given stimulus shows that the stimulus has higher sense of envelopment, localization, or movement than the reference.  However, in order to state that the difference between the stimulus and the reference is significant, statistical tests such as analysis of variance and t-test are often used. In our study, a “rule of thumb” [9] was employed to simplify the statistical inference.  Namely, in comparing average ratings of two stimuli, the two averages are statistically significantly different (at alpha=. 05 level) if the average value of a stimulus is not within the 95% confidence interval about the average value of another stimulus. Also, a stimulus is perceived significantly different from the reference on a given rating scale if the 95% confidence interval about the average rating of the stimulus does not include 3.0 in between the edges.

3.5 Sound Sample Presentation
Subjects were, for preliminary audition, presented with a15-second duration each of the reference sound sample and seven sound samples of Pattern 1 to Pattern 7 with 5-second intervals in between. Then during the actual evaluation test, subjects listened to the reference sound sample, which was recorded in compliance with the ITU-R loudspeaker placement, and subsequently to randomly present 30-sec durations of 7 patterns of sound samples with 5-sec intervals in between. To check for a correct evaluation, subjects were able to listen to each presentation as many times as they wanted.

3.6 Results
The result of evaluation by 25 subjects of 7 loudspeaker placement patterns is shown in Fig.4.




                                                                  
Fig.4 Results of Listening Test

3.7 Discussion

Following are the review of results of the evaluation listening test for the seven (7) -loudspeaker placements
patterns:

Pattern 1:
Evaluation score for Listener Envelopment (Sensation of being embraced by sounds) is as low as 2.0. This is ascribed to the weaker sound link between the front and rear loudspeakers, which are at too great a distance from one another.

Pattern 2:
The result shows that rear loudspeakers can be installed higher than the listening point without inferior aural impressions. Greater heights of rear loudspeakers do not lead to lower evaluation scores. The allowable height and vertical angle slant for rear loudspeakers, as proposed by the production side, should be within 30% of the reference height and within 15degrees of the horizontal position, respectively.

Pattern 3:
For the three evaluation attributes, good evaluation scores are obtained.

Pattern 4:
Lower evaluation score is observed for sound image localization, and sensation of movement. This can be ascribed to the rear sound field, which is rather narrow and small compared with the reference sound field.

Pattern 5:
Lower evaluation score is marked for sound image localization.
This is due to the stronger reflections from the floor.

Pattern 6:
Evaluation is low for sensation of movement. It is estimated that sense of smooth movement is difficult to
achieve because sound images move at a slanted angle.

Pattern 7:
Evaluation score is low for sound image localization. This could be ascribed to the greater deviation of loudspeaker directivity axis from the listening position.

For Patterns 5, 6, and 7, loudspeaker directivity axis orientation must also be taken into consideration.

3.8 Supplementary Experiments

As described in the above, with Patterns 5 and 7, the influence of the deviation of loudspeaker directivity axis orientation is not negligible on the perceived sound at the listening position. To study in detail the influence of deviated directivity axis, supplementary experiments were made for Patterns 5 and 7, with the directivity axis orientation varied. Evaluation was made on five (5) loudspeaker placement patterns shown in Table 2.

Patterns 1, 2, 3
All loudspeakers are placed as high as the ceiling. This placement corresponds to a case in a home-theater where all loudspeakers are mounted recessed in the ceiling.

Patterns 4, 5
Front loudspeakers are positioned closer to the floor, with rear loudspeakers closer to the ceiling. This placement allows for a case where a “rack theater” system is used.

The distance of each loudspeaker from the listening position is 2000 mm.
The angle between the center and front L or R loudspeakers is 30°, between the center and rear L or R loudspeaker 120°. The reference placement was in accordance with the ITU-R recommendation.

Pattern
Height F/C/R (mm)
TiltAngle F/C/R(°)
1
2300/2300/2300
25/25/25
2
2365/2365/2365
0/0/0
3
2220/2200/2245
90/90/90
4
460/460/2245
0/0/90
5
530/530/2245
12/12/90

Table 2.  Supplemental Loudspeaker Placement Patterns

3.9 Results of Supplemental Listening Tests
Results of supplemental listening tests are shown in Fig. 5.

 

                 
                  Fig.5 Results of Supplemental Listening Tests

Patterns 1,2, and 3
With front and rear loudspeakers placed in the same setting in terms of height, angle of separation, and directivity orientation, the aural impression is not greatly different from that obtained with the reference ITU-R placement. With Pattern 1, where directivity axis orientation is right toward the listening position, the aural impression comes closer to that of the reference ITU-R placement. With the directivity axis oriented right toward the floor of the room, the evaluation scale is slightly lower than 3.0, but with small variation in evaluation for the three evaluation attributes, suggesting evaluation is little relevant to what contents is reproduced.

Patterns 4 and 5
Patterns 4 and 5 correspond to the cases where rear loudspeakers are installed in the ceiling, with the directivity axis oriented right down to the floor. Front loudspeakers were placed, with tilt angle at 0 (Pattern 4) and 12 degrees (Pattern 5). As seen in Fig.5, the tilted placement of Pattern 5 gives better evaluation than Pattern 4 in terms of “Listener Envelopment (Sensation of being embraced by sounds)” and  “Sound Image Localization.”

4.  Conclusions

1. Cluster analysis of samples of surround loudspeaker placement in home theaters resulted in deriving five typical placement patterns.

2. Since more and more home theaters are expected to have surround loudspeakers installed in the ceiling, two placement patterns were added to the five patterns for evaluation test.
Listening evaluation was made of seven loudspeaker placement patterns against the ITU-R recommendation placement pattern for (1) Listener Envelopment (Sensation of being embraced by sounds), (2) Sound Image Localization, and (3) Sensation of movement.

The results are:
(1) Too wide angles between rear loudspeakers produce inferior Listener Envelopment (Sensation of being embraced by sounds.)
(2) With only rear loudspeakers installed in the ceiling, evaluation is lower for Sensation of movement.
(3) With all the loudspeakers installed in the ceiling, evaluation is lower for Sound Image Localization.
(4) With rear loudspeakers installed higher than front loudspeakers , or the center loudspeaker installed lower than front L and R loudspeakersthe evaluation score is comparable to that with the reference (ITU-R recommendation) placement pattern.

The results of supplemental experiments support the estimation obtained from the preliminary experiments.

The results of the listening experiments have led to the proposal of the following guidelines for surround loudspeaker placement in home theaters.

1. Angle between Rear Loudspeakers L, and R
A recommended angle is between 100 degrees and 135 degrees. The angle allows a good surround field without listener envelopment and sound image localization lost.
(ITU-R recommends 110±10 degrees.)

2. Height of Front Center Loudspeaker
If the three front loudspeakers cannot be placed at the same height, the center loudspeaker is allowed to be placed lower than L and R loudspeakers. The vertical slant angle from L and R should be within 20 degrees. The ITU-R recommends the placement at the same height as the listening position.

3. Slant Angles of Loudspeakers Installed in Wall
For loudspeakers installed higher in the walls, the upward slant angle should be within 20 degrees. The ITU-R recommends a degree within 15 degrees.

4. Distance of Rear Loudspeakers from Listening Position Relative to Front Loudspeakers.
The distance of rear loudspeakers from the listening position should be ±10% of the distance of front      loudspeakers. (The ITU-R recommends the equal
distance from the listening position for each loudspeaker.)

5. Sound Field Compensation Tool Utilization
Proper use of automatic sound field compensation features built in AV amplifiers combined with good loudspeaker placement allow general users to enjoy the surround sounds as originally intended at production.

ACKNOWLEDGEMENTS
A working group in the Digital Home-theater Committee of the Japan Audio Society carried out the research and investigation by the listening tests. It has been worked from 2010-2013.

Deep gratitude goes to those who provided the loudspeaker placement information, offered the experiment equipment and facilities, and participated in the experiments as subjects.

REFERENCES
[1] Rec. ITU-R BS.775-2, Multichannel stereophonic sound system with and without accompanying picture, ITU, 2006
[2] M. Sawaguchi et al, Surround Sound Handbook, Tokyo Geidai Press, 2010
[3] M. Sawaguchi, Current Topics on Surround Sound,
Pioneer Corporation R&D No18, 2008
[4] A.Fukada, Surround Sound viewed from the perspective of playback environment, InterBEE Forum, 2006
[5] Recommendations For Surround Sound Production, NARAS, 2004
[6] S.Yoshikawa et al, Proposal for the Specification of Control Rooms for HDTV Multichannel Sound Program Production, 100th AESConvention, 1996
[7] H. Suzuki et al, Study on Optimum Rear Loudspeaker Height for 3-1 Reproduction HDTV Audio, 95th AES Convention, 1993
[8]Olive Sean et al., Interaction between Loudspeakers and Room Acoustics Influences Loud Speakers
Preferences in Multichannel Audio Reproduction, 123rd AES Convention, 2007
[9]Geoff Cumming and Sue Finch. "Inference by Eye: Confidence Intervals and How to Read Pictures of Data" American Psychologist. 2005.


BIO of WG Member



Currently, He is funder of Hi-Res Music Label UNAMAS to product Hi-Res surround/stereo music. http://unamas-label.net/.

Mick (Masaki) Sawaguchi graduated in 1971 in electronics from the Chiba Institute of Technology and joined the Broadcasting Engineering Department of NHK (Japan Broadcasting Corporation) as an engineer. By the 80’s, he was already a highly respected figure, both within NHK and in the audio engineering industry. He was an officer of the AES Japan section until 2010. Currently, he is president of Mick Sound Lab, which is a HD surround sound production. Also, he has a surround sound design class at TOKYO UNIVERCITY of the Arts and Nagoya University of the Arts. He has had many books on surround sound production published in Japan, China and Korea. He has contributed to many AES Conferences by presenting several technical papers on surround sound mixing and microphone techniques since 1991. He is a Vice-Chairman of the AES Studio Practices and Production Committee. In 2002 he was awarded an AES Fellowship Award for his outstanding contributions to the fields of surround sound recording and mixing for broadcasting. Other awards he has been honored to receive include: ABU Best Paper at 2004 as surround broadcasting program production, a Surround Master Award by JAS, 2005, an AES JAPAN Award for his Contribution of Surround Terakoya activity, 2009, a JAS Award for a study of evaluation of home theater surround, 2012. He has chaired or participated as panelist in several workshops or sessions at AES Conventions/Conferences, SMPTE Conferences, IBS/IBC Conferences, and INTERBEE. He has also delivered lectures on surround sound recording in countries all over the world.
           

Shinji Koyano graduate from Electronics Engineering Department of Tokyo Denki University in 1975, and joined Pioneer Corporation. At the Loudspeaker Engineering Department of the corporation he designed and developed loudspeaker systems. In 1993 he got engaged in research on loudspeaker system and audio signal processing as manager of Acoustic Research Section of the Corporate Research and Development Department. Since 2005, he has been engaged in activities related to education and promotion of audio engineering, in and out of Pioneer. He is a member of Acoustic Transducer Standards Group, JEITA (Japan Electronics and Information Technology Industries Association), the head of the Surround Sound Working Group, JAS (Japan Audio Society), the secretary of 1-bit Audio Consortium.

His contribution in activity to AES includes: the Paper Chair in the 11th AES Tokyo Convention, 2003, the Paper Chair in the AES Japan Conference, 2006, the Convention Chair of the 13th AES Tokyo Convention, and the Chair of the AES Japan Section from 2007 to 2008. He was awarded a Board of Governors Award in 2009 for his successful presiding over 13th AES Tokyo Convention. He is also a member of ASJ (Acoustical Society of Japan).






Masami “Sam” Toyoshima graduated from Waseda University in 1964 with a Master Degree of Engineering. On graduation, he joined JVC’s Audio Lab. At JVC in 1977 he started designing recording studios, along with his colleague Hiroaki “Bike” Suzuki. In addition to JVC’s Studio, he designed major studios throughout Japan including Warner Music, Onkio Haus, Roland, On Air, West Side, etc. Meanwhile, in UK in 1985, he established Acoustics Design Group with an architect, John Flynn, Bike Suzuki and a producer/engineer, Hugh
Padgham. He designed major studios in London, including Abbey Road, Town House, Olympic, Metropolis, Sarm West, and so on. He has done more than 250 music-recording facilities worldwide. From 2001 he has been teaching acoustics, as Professor at Yokkaichi University. He received an AES Fellowship Award in 2011 in London. He is a member of the AES, Acoustic Society of Japan, Institute of Electronics, Information and Communication Engineers, The Virtual Reality Society of Japan, and is an advisor to the Japan APRS. He presented some technical papers at the past AES Conventions and Conferences.



Kazuhiko Terui joined Sony Corporation in 1978. He was engaged in designing AVsurround amplifiers and tuners at the Audio Component Design Division. In 1990, he was transferred temporarily to CBS Sony Corporation to have some experience in the field of music production through working in the Japanese music division. After getting back to Sony, he worked in the Product Promotion Division, Overseas Sales Division, then was deeply involved in the strategy of promoting SACD, and worked aggressively to enrich the list of SACD titles. Currently he is the Chairman of the Audio Network Business Committee, JEITA (Japan Electronics and Information Technology Industries Association).




 Bike Hiroaki Suzuki graduated from Tohoku University with a bachelor degree in electrical engineering in 1970, and joined JVC to get involved in the field of architectural acoustic design. Design projects he worked on with his boss, Sam Toyoshima, includes Townhouse Studio, Abbey Road Studios, and Metropolis Studios which are all located in London. In 1997 he was appointed as Chairman of WG-4 of DVD Forum, which was responsible for producing the new audio format, DVD-Audio Specifications. In 2001 he was awarded an AES Fellowship Award in recognition of continued contribution to acoustics and studio design as well as his leading role in the development of DVD-Audio.