Sunday, November 22, 2009

The Effect of Whole-body Vibrations on Preferred Bass Equalizations of Automotive Audio Systems

Binaural Room Scanning (BRS) is a technology that allows Harman scientists to binaurally capture, store, and later reproduce sound fields through a headphone-based auditory display that includes head tracking for accurate localization of sound sources [1]. BRS enables us to do controlled, double blind comparative evaluations of different automotive audio systems, home theatre systems or sound reinforcement systems that would otherwise not be practical or possible to do. However, current BRS systems do not typically capture and reproduce the whole-body vibrations that are associated with low frequencies reproduced by the audio system. Therefore, an important question related to the accuracy and ecological validity of BRS-based evaluations is whether whole-body vibrations play an important role in our perception of the quality and realism of the automotive audio system.


We recently presented a paper at the 127th Audio Engineering Society Convention that addressed this research question [2]. Three experiments were reported that measured the effects of both real and simulated whole-body vibration associated with the low frequency sounds reproduced by an automotive audio system on listeners’ preferred bass equalizations. A PDF of the side presentation can be found here.


In all three experiments, the same automotive audio system was used: a high-quality 17-channel audio system installed in a 2004 Toyota Avalon. The car was parked in our automotive audio research lab with the engine turned off so that the effects of the road and engine noise and vibration were not part of the experiment. The BRS system was calibrated for each individual listener to minimize errors related to headphone fit, etc. The differences in magnitude response measured at the listeners’ ear in situ versus through the BRS system were very small indeed (see slide 9).


The Effect of Real Vibration on Preferred Bass Levels


In the first experiment, listeners adjusted the level of bass equalization (see slide 10) while experiencing real whole-body vibration produced by the car audio system. The same task was also repeated via the BRS headphone-based system without the vibration present. This was repeated three times using three different music programs (see slide 7). On average, listeners adjusted the bass equalization 1.5 dB higher for the BRS playback condition where the low frequency vibration was not present (see slide 11). The preferred bass level was found to be program dependent due to the amount of bass present in the signal, and the resulting vibration it produced(see slides 12 and 13).


The Effect of Simulated Vibration on Preferred Bass Levels


In the second experiment, we simulated the whole-body vibration produced by the audio system by attaching an actuator to the driver's seat of the car. The actuator was driven by the low frequency portion of the audio signal below 100 Hz. Comprehensive whole-body vibration measurements performed prior to this experiment found that most of the whole-body vibration produced by the audio system occurs below 100 Hz (see slides 15 and 16) at the seat and floor. The level and frequency of the whole-body vibration varies with music program, and the weight of the listener.


Each listener sat in the driver’s seat of the car listening a virtual BRS rendering of the automotive audio system reproduced through headphones. Listeners adjusted the bass level in their headphones while experiencing four different levels of simulated whole-body vibration that varied from none, low (0 dB) medium (+4 dB) and high (+8 dB). The medium level corresponded to the measured vibration in experiment one, when the bass equalization of the automotive audio system was adjusted to its preferred level.


The experimental results indicated that the preferred level (dB) of bass equalization decreased 3 dB when the level of whole-body vibration increased 8 dB. (see slide 21), and varied with program. At the low vibration level, there was no effect on the preferred bass level, since the vibration level was near or below detection thresholds reported in the literature. At the highest level (8 dB), the vibration tended to be annoying, and listeners tended to turn the bass level down with the hope that the vibration would also be reduced. The effect of vibration on preferred bass level was somewhat dependent on the listener, which could be related to their weight (see slide 24).


The experimental results confirm those of a previous experiment conducted by Martens et al. where the vibration was simulated via a platform, and both head-tracking and individualized BRS calibrations were not employed [3]. The results from the Martens' et al. experiments and this one are plotted above in Figure 1. In spite of the methodological differences between the two experiments, there is good agreement between the two studies. This suggests that the effect of whole-body vibration on the preferred level of bass equalization is quite robust.


The Effect of Whole-body Vibration on the Similarity in Sound Quality between BRS and In Situ Reproductions


The third experiment, listeners sat in the car and rated the overall similarity in sound quality between the BRS headphone-based reproduction with and without the simulated vibration compared to the same audio system experienced in situ. Listeners could switch at will between the in situ and two BRS reproductions (with and without shaker). The two BRS treatments were presented double-blind, and repeated two times with four music programs (see slide 27).


The results (see slide 29) show that sound quality of the BRS reproduction system was significantly improved with the presence of whole-body vibration (shaker on).


Conclusions


From these experiments, it is clear that the whole-body vibration associated with the low frequency sounds of an audio system influences listeners’ perception of the quality and quantity of bass. When the vibration is absent from a stereo or binaural recording of music reproduced through headphones there may be a perceived lack of bass. A 4 dB increase in whole-body vibration produces about a 1.5 dB decrease in preferred level of bass equalization. However, there appears to be upper and lower threshold limits beyond which a change in vibration level will have no effect. Moreover, the amount of vibration and its effect on preferred levels of bass equalization will depend on the low frequency characteristics of the music and the individual listener (and possibly their weight).


Finally, adding simulated whole-body vibration to BRS reproductions can greatly enhance their perceived realism and fidelity when compared to the in situ experience, as long as the vibration levels are above the listener's detection threshold.


References


[1] Sean E. Olive and Todd Welti, “Validation of a Binaural Car Scanning Measurement System for Subjective Evaluation of Automotive Audio Systems,” presented at the 36th International AES Automotive Audio Conference, (June 2-4, 2009).


[2] Germain Simon, Sean E. Olive, and Todd Welti, “The Effect of Whole-body Vibration on Preferred Bass Equalization in Automotive Audio Systems,” presented at the 127th Audio Eng. Soc. Convention, preprint 7956, (October 2009).



[3] William Martens, Wieslaw Woszczyk, Hideki Sakanashi, and Sean E. Olive, “Whole-Body Vibration Associated with Low-Frequency Audio Reproduction Influences Preferred Vibration,” presented at the AES 36th International Conference, Dearborn, Michigan (June 2-4, 2009).


11 comments:

  1. Please check the link to the presentation, google says not available. Grant

    ReplyDelete
  2. Thanks for the heads up. It should be shared to everyone now.

    ReplyDelete
  3. Dear Dr.Olive,

    What kind of solution would you propose to compensate this bass-lacking issue in headphones due to the absence of whole-body vibration? Wouldn't simply amplifying low frequency just mask the upper frequencies, and might introduce an additional phase-related issue?

    How about Harman Kardon comes up with a new headphone with a whole-body vibrator as an addon? (I hope to see Harman Kardon collaborating with Etymotic Research once again, as you guys did with EP-730, to resolve this MISSING-6dB)

    Thank you,
    udauda.

    ReplyDelete
  4. I have 2 friends with theaters on "suspended" floors. Their subwoofer systems are able to "excite" their floors to vibrate. The effect is quite palable, and is clearly missing in my HT, which is in my basement on a concrete floor.

    Frankly, I'm surprised that you *only* measured a 1.5 dB difference in preference. I would have thought that the difference was at least 6 to 10 dB.

    I have a Crowson Tactile Transducer under my seating to try to mimic the vibration experienced in my friends' suspended theaters. It comes close, but it's not the same.

    Craig

    ReplyDelete
  5. Motor off?
    Car not moving?
    Front passenger using headphones?
    Tactile transducer attached to seat?
    .
    These conditions are so far from real world listening to music in a moving car, that I suspect the conclusions are meaningless except for people who listen in a car with the motor off, not moving, wearing headphones, and having fun with a new toy (tactile transducer)!

    The listening conditions in any audio test should simulate typical listening conditions by typical listeners, assuming you want meaningful conclusions from the test.
    .
    The experiment did demonstrate that bass is both heard and felt. Audiophiles know body pressurization becomes more important as the bass frequency declines ... and hearing efficiency fades. So if you listen to headphones, and only "feel" bass on your eardrums, the bass will seem weak even though the SPL measurement doesn't say that.
    .
    I wonder what was gained from an experiment that "proved" the almost universal belief (to everyone who has ever used headphones) that bass is not as much fun through headphones!

    Richard Greene
    richgreene2@yahoo.com

    ReplyDelete
  6. Hi Richard,

    If you read our AES paper we point out that the presence of noise and vibration from the road/engine could swamp out any effects reported in this paper. More experiments would be needed to verify that, and such tests are planned in the future.

    However, doing evaluations of different car audio systems under road conditions has its own set of logistical and scientific challenges and problems. How do you control the effects of the road noise/vibration from the other effects you are trying to measure? What about repeatability? How would you compare two different car audio systems when driving them?

    The solution is to capture, store and reproduce the car audio systems using a BRS system like the one we developed, and then add the real or simulated noise and vibration from the road/engine. That is why we developed the BRS system. And that is why it uses headphones. The question this experiment addressed was "Do you perceptually miss something when using headphones?"

    This experiment was designed to study the effects of vibration produced by the audio system -- not the vibration from the road/engine. That doesn't mean the experiments didn't produce meaningful results. I can think of many real-life situations where the car is stationary ( a car show room, LA traffic, parked car waiting for your kids) without the engine running. With the popularity of hybrid and electric vehicles car engines are very quiet).

    If we included the road/engine vibration before measuring the effect of just the audio system we could not separate out the cause of the effect. This is just good scientific practice of separating and controlling variables so you understand the underlying relationship between cause and effect.

    Scientific proof and "a universal belief" are not the same thing. Many people believed the earth was flat until someone proved they were wrong. This experiment provided some experimental evidence that the vibration produced from an audio system cannot be ignored when you are trying to capture and evaluate the audio system through a binaural system. We also found that the real effect can be simulated, how to calibrate the effect, and that there are upper and lower limits beyond which we no longer detect or accept the plausibility of vibration.

    For me, that is useful knowledge gained from this study.

    ReplyDelete
  7. Good vibrations-analysis will surely give you an accurate output on the test you're conducting.

    ReplyDelete
  8. You has done a great work. I appreciate your work.I am highly oblige for this because it helps me a lot.

    ReplyDelete
  9. Whole Body Vibration is one of the most efficient ways to exercise because it fatigues the muscles quickly and most effectively with minimal effort from the user.and it can be used in the comfort of your own home. Whole Body Vibration

    ReplyDelete
  10. I surely agree that whole body vibration associated with the low frequency sounds of an audio system influences listeners’ perception of the quality and quantity of bass. Thanks for sharing

    ReplyDelete