Professional Documents
Culture Documents
Convention Paper: In-Vehicle Audio System Sound Quality Preference Study
Convention Paper: In-Vehicle Audio System Sound Quality Preference Study
This paper was peer-reviewed as a complete manuscript for presentation at this Convention. This paper is available in the AES
E-Library, http://www.aes.org/e-lib. All rights reserved. Reproduction of this paper, or any portion thereof, is not permitted
without direct permission from the Journal of the Audio Engineering Society.
ABSTRACT
In-vehicle audio systems present a unique listening environment. Listeners were asked to adjust the relative bass
and treble levels as well as fade and balance levels based on preference on three music programs reproduced through
a high quality in-vehicle audio system. The audio system frequency response was initially tuned similar to that
preferred for in-room loudspeakers. The fade control was initially set to give a frontal image with some rear
envelopment using two different rear speaker locations, rear deck and rear door, while the balance control was set to
give a center image between the center of the steering wheel and rearview mirror. Stage height was located on top of
the instrument panel (near head level). Results showed that on average listeners preferred +13 dB bass and -2 dB
treble compared to a flat response while fade was +3.5 dB rearward for rear deck mounted speakers,+2.6 dB
rearward for rear door mounted and balance was 0 dB. Significant variations between individual listeners were
observed.
Evidence exists that suggests listeners do have differing Further studies will be needed to investigate how trim
preferences based on playback method. Olive et al [3] panel/sheet metal resonances, noise dampening, vehicle
investigated listener frequency response preferences for noise during driving and other characteristics affect
in-room loudspeakers and headphones and found that on sound quality preference.
average listeners preferred 2dB more bass and treble for
in-room loudspeakers when compared to headphones. 2.2 Audio Equipment
This paper describes a set of listening experiments to The following audio equipment was installed into the
help gain understanding of listener preferences in a vehicle [5]:
vehicle environment. 8 Expert and 95 non-Expert
listeners were asked to adjust the following on a high
quality in-vehicle audio system for 3 music tracks: -Roland Octa-Capture
-Rockford Fosgate 360.3 Signal Processor (x1)
-Rockford Fosgate Power Series T3-BMW3
- Bass level only, treble fixed at 0dB (-3dB in (x2 sets) 4” component speaker set
reference to flat) -Rockford Fosgate Power Series T-165-S (x2
- Treble level only, bass fixed at 0dB (+10dB in sets) 6.5” component speaker set
reference to flat) -Rockford Fosgate Power Series T-1692 (x1
- Bass and treble level together set) 6x9 speakers
- Fade level with rear speakers in the rear deck -Rockford Fosgate Punch Series PL3-112 12”
- Fade level with rear speakers in the rear doors Subwoofer in ported enclosure
- Balance level -Rockford Fosgate Power Series T600-2 (x1)
Power Amplifier
-Rockford Fosgate Power Series T400-4 (x3)
2. EXPERIMENTS ON PREFERRED LEVELS Power Amplifier
(BASS, TREBLE, BALANCE AND FADE) -KnuKonceptz Power wire (1/0 ga & 4 ga)
-KnuKonceptz Speaker wire (12 ga)
This section describes the test methodology used for this -Audio stereo switching box
study. This includes details of the vehicle set up,
equalization method, test procedure, program material 2.3 Audio System Tuning
and listeners.
The audio system equalization approach was to
2.1. Vehicle Set-up optimize each amplifier channel/speaker combination
individually first at the driver head location and then
An aftermarket 13 speaker Rockford Fosgate audio incorporate the channels together, optimizing at each
system was installed into a 2015 Nissan Altima. OEM stage.
speaker locations were utilized along with some custom
locations for the front and rear tweeters as well as the First the impulse responses were measured through each
trunk for the subwoofer [4] (see Appendix for amplifier channel to the driver’s head location. Each
locations). The vehicle doors (inner and outer sheet channel was then time aligned to the driver location.
metal) as well as rear deck were dampened with a sound The rear channel measurements (rear deck or rear door
deadening material to reduce unwanted resonances. The depending on configuration) were averaged and a
stock plastic trim grills were removed and replaced with common delay used for left rear and right rear.
a metal or cloth based grill for better transparency.
Acoustic measurements for tuning were then performed
The vehicle modifications were done to achieve the best using a spatial average of 6 omnidirectional
sound quality possible so that the in-vehicle listening microphones [6]. The target response was chosen to be
space was the dominant effect on the sound quality. similar to the preferred in-home loudspeaker response,
Because interior materials and quality are quite different with an additional 4 dB of bass gain (+10 dB total in
from an entry level vehicle to a luxury vehicle we were reference to flat) which made the audio system sound
most interested in the physical space without the more natural [7], see Fig. 1. Instead of flattening the
complications of the other aspects of a typical in-vehicle response from 200 Hz to 1.5 kHz, a downward slope
audio system which involve trade-offs in sound design.
Figure 1 - Preferred In-Room response from Olive et. al. Figure 4 - Measured In-Vehicle 1/12 Octave response
[3] vs In-Vehicle initial response faded full front vs faded full rear, Rear Door speakers
When the listener adjusted the bass or treble control, the 2.6 Fade Control
test program would pick the appropriate signal, i.e. if
the listener wanted +3 dB of bass and – 3 dB of treble, The fade curve was constructed so that the loudness and
the program would mix the +3 dB low-pass signal, the frequency response of the material stayed relatively
inverted band-pass signal and the -3 dB high-pass signal constant regardless of the control position. The low-
together. Fig. 6 shows the response difference between passed signal was not controlled by the fader, only the
the original signal and the original signal reconstructed band-pass and high-passed signals in order to keep the
from the 3 Linkwitz-Riley filtered signals. bass level and overall response as constant as possible.
The initial setting was randomized Front/Rear:
-0.5dB/2.5dB to 1dB/-2.5dB
order was bass/treble tests first, fade/balance tests last. 2.10 Program Selections
We were most interested in the bass and treble level
settings and added the fade/balance test as additional The music selections were the same used in Olive et al
data for collection. [3] so that data from this preference study could be
loosely compared.
Program/Artist/Track/Album Description
JW - Jennifer Warnes/Bird on a
Female Pop
Wire/Famous Blue Raincoat/
Vocal
Cypress Records, 1986
SD - Steely Dan/Cousin
Male Pop
Dupree/Two Against Nature/
Vocal
Giant Records/ 2000
ES - Estelle w. Kanye West/
Male/Female
American Boy Shine/
Figure 7 - Test GUI used by Test Administrator Hip Hop
Shine/Atlantic Records, 2008
Table 1 - Details on the music selections used for these
As in Olive et al [3] three Griffin Powermate [10] USB listening experiments
assignable controllers were used, one for bass, one for
treble and one for fade/balance/volume. The controls The programs were digitally transferred from compact
are continuously rotary with no detents so as not to bias disc and edited into 20 – 30 s loops. Fig. 8 shows the
the listener with visual or tactile feedback related to the average long-term spectrum of the three music loops
position of the knob. No dead zones were incorporated indicating they were sufficiently spectrally dense and
into the knobs for ease of test facilitation. The broadband for this test.
sensitivities of the knobs were set to minimum for bass
and treble, and ¾ for fade. These sensitivity settings
along with loudness compensation made it difficult to
identify the control extremes.
Figure 9 - Participant experience levels So why the difference in preferred bass level in
vehicles? This amount of gain is consistent with
automotive audio systems regarded as having good
sound [7]. The space in a vehicle is quite small,
resulting in a reduced number of modes < 150 Hz
compared to a typical small room. Also, reverberation
times are less, leaving the main source of bass energy as
direct sound. It possibly could be that the lack of modes
and reverberant reinforcement caused a psychoacoustic
effect of the audio system lacking preferred bass at
lower gain levels.
3.3 Two Parameter MOA Spatial Results It was surprising there was no significant difference
measured between Rear Deck location vs Rear Door.
MANOVA and ANOVA were performed for each two With the Rear Deck speaker location the sound stage
parameter spatial test. For the fade test the fixed height was preserved near head level, while the rear
independent variables were Playback Method (2 levels: door location pulled the stage height down, more so the
Rear Deck location and Rear Door location), Program further rear the participant faded.
(3 levels) and Listener Experience (2 levels: Expert and
non-Expert). The dependent variables were preferred For the balance position, it appears as if the listeners on
volume and fade position. The balance test had the average are more interested in a balanced sound stage.
same fixed independent variables, dependent variables The preferred position put the center image between the
were preferred volume and balance position. steering wheel and rearview mirror with a relatively
No significant factors where found. The values in the uniform stage distribution from left A-pillar to right A-
following graphs were calculated as left minus right for pillar instead of having the center image directly in front
balance, front minus rear for fade. of them which shortened the stage on the driver side and
expanded it towards the passenger side. This was true
Fig. 21 shows Balance/Fade comparison by Experience for Experts and Non-Experts alike.
and Fig. 22 by Volume. It appears that the preference
for the sound is to be more enveloping than the Fig. 23 Shows levels preferred between programs and
reference sound of the performance in front of the Fig. 24 shows levels preferred between gender.
listener and ambience cues from the rear. This desired
amount of level may differ depending on the signals
delivered to the rear speakers, ambience only vs
duplicating the front channel signals. Further study will
be needed.
4 OBJECTIVE MEASUREMENTS OF
PREFERRED TARGET RESPONSES
Fig 25 plots the preferred bass and treble for Bass Only
Treble Only test (+3.2 dB, +0.9 dB respectively) as well
as for the Bass and Treble together test (+3.2dB, +2.1dB
respectively).
5 CONCLUSIONS
For the Fade tests the initial setting was set to a frontal [6] Geddes E., Blind H., “The Localized Sound Power
stage, height at top of instrumental panel (near head Method”, JAES Vol 34, Issue 3, March 1986
height) with some ambience contribution from the rear
speakers. Listeners tended to fade the control further [7] Rumsey F., “Orchestrating Automotive Audio –
rearward (approximately 3 dB), increasing the Uneven Bass Reproduction in Automobiles”, JAES
envelopment feeling provided by the rear speakers. The Vol 56, Issue 4, April 2008
sensation was more preference then reference, a less
dominant frontal image with noticeable rear [8] Olive S.E., Jackson J., Devantier A., Hunt D. and
contribution. Future work is needed to study this effect Hess S., “The Subjective and Objective Evaluation
with different rear speaker signals, such as ambience of Room Correction Products”, 127th Convention,
information only. Audio Eng. Soc., preprint no. 7960, New York,
(2009, October)
The starting location for the center image was centered
between the steering wheel and rear view mirror. The [9] ITU BS.1770-3 Algorithms to Measure Audio
listeners on average did not move the image from this Programme Loudness and True Peak Audio Level
location. It appears they are looking for a wide, evenly
spaced stage instead of one that tends to be narrowly [10] Griffin Powermate 3.0,
focused and directly in front of them. http://store.griffintechnology.com/laptops/powerma
te (July 2013)
6 ACKNOWLEDGEMENTS [11] Harman How to Listen: A Listener Training
Software Program for Mac and Windows,
The author would like to thank Nissan who supported www.harmanhowtolisten.blogspot.com (March
this work, Alex Nelson of Kettering University whose 2013)
diligence made this project a success and Rockford
Fosgate whose resources and experience were [12] Olive S. E., “Differences in Performance and
invaluable. Preference of Trained versus Untrained Listeners in
Loudspeaker Tests: A Case Study” JAES Vol 51,
7 REFERENCES Issue 9, September 2003
[1] Shively R., “Automotive Audio Design (A [13] Davis M., “Automotive Audio – Equalization of
Tutorial)” 109th Convention, Audio Eng. Soc., Car Audio Systems”, JAES Vol 51, Issue 6, June
preprint no 5276, Los Angeles (2000 September) 2003
[4] Shively R., Halley J., Malbos F., Ruiz G., “Optimal
Location and orientation for Midrange and High
Frequency Loudspeakers in the Instrument Panel of
an Automotive Interior” 129th Convention, Audio
Eng. Soc., preprint no.8249, San Francisco, (2010
November)
[5] www.rockfordfosgate.com
8 APPENDIX
System Diagram
T-400 - 4
T-400 - 4
T-400 - 4
T-600 - 2