“Audio Science in the service of art” is a philosophy in sound reproduction where the goal is to use science and technology to faithfully reproduce the art as the musical artist intended [1]. The art is the music, its performance, and the process of capturing it on the recording. The audio system is not part of the art, and should neither add, remove or editorialize the artist’s message. Audio components should not sound like musical instruments: Beethoven never wrote parts for loudspeaker and amplifier, so you shouldn’t be hearing them when listening to recordings of his music. The perfect audio system has no sonic personality, no musical qualities, and is the system that you notice the least.
Many audio companies do not subscribe to this scientific approach towards sound reproduction. Some companies will admit they simply cannot afford the infrastructure required to conduct proper scientific-based objective and subjective measurements. Anechoic chambers, dedicated listening rooms, speaker movers, listening test software, trained listening panels, and educated and trained scientific/engineering staff require significant long-term financial commitment to R&D. Other companies view sound reproduction as an artistic or marketing driven opportunity to screw with the art, and perform cosmetic surgery on the music: a little bass augmentation here, a little midrange tuck there, lift the treble here, and the facelift is complete. The problem with this approach is that recordings can be flawed in an infinite number of different ways. A cure-all bandaid solution will ultimately do more harm than good for most recordings, especially the good ones, which is a disservice to the art.
Faithful reproduction of the art assumes that the listening conditions in which the art was created are well defined. While there exists many accurate monitor loudspeakers in the marketplace today (e.g. the JBL LSR professional monitor series are designed to the same targets as our consumer loudspeakers) there is nothing that guarantees the artist will use them. Without meaningful loudspeaker and room calibration standards common to the professional and consumer audio industries, recordings and their reproduction remains a poorly controlled, highly variable process. For example, a survey of 164 professional recording studios in Europe using the same factory-calibrated 3-way monitor found up to 25 dB variations below 100 Hz measured at the mixing position. It’s no wonder the bass is so variable among different recordings!
Both the consumer and professional audio industries are trapped in a codependent relationship whereby the perceived sound quality of one’s product is interdependent on the others’. Known as the circle-of-confusion, this unfortunate state of affair can only be solved through a common standard that defines the performance of the loudspeaker and its acoustical interaction with the room.
Faithful reproduction of the art requires a thorough scientific understanding of the relationship between the perception and measurement of sound so that the important variables can be identified. Most audio scientists agree that the circle-of-confusion problem will not be solved until we optimize the performance of the loudspeaker and its acoustical interaction with the room acoustics. Our scientific understanding of what makes a loudspeaker sound accurate and neutral is already well understood.
This research question was studied at the Canadian National Research Council (NRC)[2],[3]and more recently, at Harman International [4]-[6], the parent company of loudspeaker brands Infinity, Harman Kardon, JBL and Revel. Using scientific-based, double-blind loudspeaker listening tests, scientists studied which physical parameter of loudspeaker performance were most related to listeners’ sound quality ratings, and overall preference. To eliminate the effects of sighted biases (e.g. brand, price, size, reputation) the tests were performed double-blind with other known listening test nuisance variables carefully controlled. The loudspeaker positional effects in comparative loudspeaker tests were solved by an automated speaker shuffler that positions each speaker into the exact same position.The tests were performed using trained listeners with normal hearing. More recent tests with untrained listeners indicate they also prefer the same loudspeakers as trained listeners,but give less consistent and discriminating ratings.
The results of this research found that the preferred loudspeakers in the listening tests were also the most accurate ones,based on a set of a comprehensive anechoic measurements. The measurements used high frequency resolution (48 points per octave), and employed spatial averaging to separate resonances from diffraction/acoustic interference effects. The frequency response curves were then spatially averaged into a family of curves, based on a survey of user's set ups in their rooms, rooms that represent the quality of the direct, early and late reflected sounds heard in the room. A mathematical preference model based on these measurements has been recently developed and can predict the loudspeaker preference with a correlation of r = 0.86 (the agreement between the predicted and measured ratings of 70 different loudspeakers). The model tells us that both the quality of the direct and reflected sounds produced by the loudspeaker are almost equally as important, and the bass quality accounts for about 30% of a listeners’ loudspeaker preference. This suggests that the low frequency interaction between the loudspeaker and room acoustics is something that cannot be ignored.
The acoustical interaction between the loudspeaker and the room is the remaining problem that must be solved to close the loop between the creation and reproduction of the art. At low frequencies (below 200-300 Hz), all listening rooms contain a natural set of resonances or room modes that can significantly boost and attenuate low frequencies below 200-300 Hz [7]. The level and frequency of these resonances will depend on the room’s dimensions, geometry and absorption characteristics, as well as the locations of the loudspeakers and listeners.
Fortunately, there are solutions today that can deal with these low frequency variations that occur between the loudspeaker and its acoustical interaction with the room. Bass in rooms can be tamed by judiciously placing the loudspeakers and listeners in locations where the room modes have the least effect [8]. In rectangular rooms, placing multiple (2 to 4) subwoofers in the room’s corners, wall midpoints,or at 25% and 75% along the wall dimension can cancel order modes via constructive interference, and not excite others. This solution has the benefit of reducing the spatial variance in bass quality across the listening area. Finally, equalization at single or multiple seating locations avoid exiting othersreduce some of the most deleterious effects. However, not all commercial room correction solutions are equal: some models can actually make the audio system sound worse than without correction.
In summary, our scientific understanding of the relationship between the measurement and perception of loudspeakers and rooms is quite mature. Measurements exist today that can accurately and reliably predict loudspeaker sound quality, and there are practical and effective solutions for dealing with their acoustical interaction with listening rooms at low frequencies.
It's time for the audio industry to finally close the loop between the recording and playback chains - to break out of the circle of confusion. A meaningful standard that defines the performance of the playback chain where the art is both created and reproduced would certainly be good place to start. Work on a new loudspeaker standard based on the NRC and Harman loudspeaker measurements is underway within the CEA and CEDIA standards groups. When completed, consumers will have access to product specifications that identify the excellent loudspeakers from the ones that are duds. Hopefully, the professional audio industry will adopt a similar standard so that consumers hear the music as it was intended by the artist. That would be the ultimate reward for audio science in the service of art.
References
[1] Floyd E. Toole. "Science in the service of art", Harman International white paper.
[2] Floyd E. Toole, "Loudspeaker Measurements and Their Relationship to Listener Preferences: Part 1" J. AES Vol. 23, issue 4, pp. 227-235, April 1986. (download for free courtesy of Harman International).
[3]Floyd E. Toole, "Loudspeaker Measurements and Their Relationship to Listener Preferences: Part 2," J. AES, Vol. 34, Issue 5, pp. 323-248, May 1986. (download for free courtesy of Harman International).
[4] Sean E. Olive, "Differences in Performance and Preference of Trained Versus Untrained Listeners in Loudspeaker Tests: A Case Study," J. AES, Vol. 51, issue 9, pp. 806-825, September 2003. (download for free courtesy of Harman International).
[5} Sean E. Olive Sean E. Olive, "A Multiple Regression Model for Predicting Loudspeaker Preferences using Objective Measurements: Part 1 -Listening Test Results," presented at the 116th AES Convention, May 2004.
[6] Sean E. Olive "A Multiple Regression Model for Predicting Loudspeaker Preferences using Objective Measurements: Part 2 - Development of the Model", presented at the 117th AES Convention, October 2004.
[7] Floyd E. Toole " Loudspeakers and Room - A Scientific Review", J. Audio Eng. Soc., Vol. 54, No. 6, June 2006 (download for free here courtesy of Harman International)
[8] Todd Welti and Allan Devantier,"Low Frequency Optimization Using Multiple Subwoofers", J. Audio Eng. Soc., Vol. 54, No. 5, May 2006 (download for free, courtesy of Harman International).
Many audio companies do not subscribe to this scientific approach towards sound reproduction. Some companies will admit they simply cannot afford the infrastructure required to conduct proper scientific-based objective and subjective measurements. Anechoic chambers, dedicated listening rooms, speaker movers, listening test software, trained listening panels, and educated and trained scientific/engineering staff require significant long-term financial commitment to R&D. Other companies view sound reproduction as an artistic or marketing driven opportunity to screw with the art, and perform cosmetic surgery on the music: a little bass augmentation here, a little midrange tuck there, lift the treble here, and the facelift is complete. The problem with this approach is that recordings can be flawed in an infinite number of different ways. A cure-all bandaid solution will ultimately do more harm than good for most recordings, especially the good ones, which is a disservice to the art.
Faithful reproduction of the art assumes that the listening conditions in which the art was created are well defined. While there exists many accurate monitor loudspeakers in the marketplace today (e.g. the JBL LSR professional monitor series are designed to the same targets as our consumer loudspeakers) there is nothing that guarantees the artist will use them. Without meaningful loudspeaker and room calibration standards common to the professional and consumer audio industries, recordings and their reproduction remains a poorly controlled, highly variable process. For example, a survey of 164 professional recording studios in Europe using the same factory-calibrated 3-way monitor found up to 25 dB variations below 100 Hz measured at the mixing position. It’s no wonder the bass is so variable among different recordings!
Both the consumer and professional audio industries are trapped in a codependent relationship whereby the perceived sound quality of one’s product is interdependent on the others’. Known as the circle-of-confusion, this unfortunate state of affair can only be solved through a common standard that defines the performance of the loudspeaker and its acoustical interaction with the room.
Faithful reproduction of the art requires a thorough scientific understanding of the relationship between the perception and measurement of sound so that the important variables can be identified. Most audio scientists agree that the circle-of-confusion problem will not be solved until we optimize the performance of the loudspeaker and its acoustical interaction with the room acoustics. Our scientific understanding of what makes a loudspeaker sound accurate and neutral is already well understood.
This research question was studied at the Canadian National Research Council (NRC)[2],[3]and more recently, at Harman International [4]-[6], the parent company of loudspeaker brands Infinity, Harman Kardon, JBL and Revel. Using scientific-based, double-blind loudspeaker listening tests, scientists studied which physical parameter of loudspeaker performance were most related to listeners’ sound quality ratings, and overall preference. To eliminate the effects of sighted biases (e.g. brand, price, size, reputation) the tests were performed double-blind with other known listening test nuisance variables carefully controlled. The loudspeaker positional effects in comparative loudspeaker tests were solved by an automated speaker shuffler that positions each speaker into the exact same position.The tests were performed using trained listeners with normal hearing. More recent tests with untrained listeners indicate they also prefer the same loudspeakers as trained listeners,but give less consistent and discriminating ratings.
The results of this research found that the preferred loudspeakers in the listening tests were also the most accurate ones,based on a set of a comprehensive anechoic measurements. The measurements used high frequency resolution (48 points per octave), and employed spatial averaging to separate resonances from diffraction/acoustic interference effects. The frequency response curves were then spatially averaged into a family of curves, based on a survey of user's set ups in their rooms, rooms that represent the quality of the direct, early and late reflected sounds heard in the room. A mathematical preference model based on these measurements has been recently developed and can predict the loudspeaker preference with a correlation of r = 0.86 (the agreement between the predicted and measured ratings of 70 different loudspeakers). The model tells us that both the quality of the direct and reflected sounds produced by the loudspeaker are almost equally as important, and the bass quality accounts for about 30% of a listeners’ loudspeaker preference. This suggests that the low frequency interaction between the loudspeaker and room acoustics is something that cannot be ignored.
The acoustical interaction between the loudspeaker and the room is the remaining problem that must be solved to close the loop between the creation and reproduction of the art. At low frequencies (below 200-300 Hz), all listening rooms contain a natural set of resonances or room modes that can significantly boost and attenuate low frequencies below 200-300 Hz [7]. The level and frequency of these resonances will depend on the room’s dimensions, geometry and absorption characteristics, as well as the locations of the loudspeakers and listeners.
Fortunately, there are solutions today that can deal with these low frequency variations that occur between the loudspeaker and its acoustical interaction with the room. Bass in rooms can be tamed by judiciously placing the loudspeakers and listeners in locations where the room modes have the least effect [8]. In rectangular rooms, placing multiple (2 to 4) subwoofers in the room’s corners, wall midpoints,or at 25% and 75% along the wall dimension can cancel order modes via constructive interference, and not excite others. This solution has the benefit of reducing the spatial variance in bass quality across the listening area. Finally, equalization at single or multiple seating locations avoid exiting othersreduce some of the most deleterious effects. However, not all commercial room correction solutions are equal: some models can actually make the audio system sound worse than without correction.
In summary, our scientific understanding of the relationship between the measurement and perception of loudspeakers and rooms is quite mature. Measurements exist today that can accurately and reliably predict loudspeaker sound quality, and there are practical and effective solutions for dealing with their acoustical interaction with listening rooms at low frequencies.
It's time for the audio industry to finally close the loop between the recording and playback chains - to break out of the circle of confusion. A meaningful standard that defines the performance of the playback chain where the art is both created and reproduced would certainly be good place to start. Work on a new loudspeaker standard based on the NRC and Harman loudspeaker measurements is underway within the CEA and CEDIA standards groups. When completed, consumers will have access to product specifications that identify the excellent loudspeakers from the ones that are duds. Hopefully, the professional audio industry will adopt a similar standard so that consumers hear the music as it was intended by the artist. That would be the ultimate reward for audio science in the service of art.
References
[1] Floyd E. Toole. "Science in the service of art", Harman International white paper.
[2] Floyd E. Toole, "Loudspeaker Measurements and Their Relationship to Listener Preferences: Part 1" J. AES Vol. 23, issue 4, pp. 227-235, April 1986. (download for free courtesy of Harman International).
[3]Floyd E. Toole, "Loudspeaker Measurements and Their Relationship to Listener Preferences: Part 2," J. AES, Vol. 34, Issue 5, pp. 323-248, May 1986. (download for free courtesy of Harman International).
[4] Sean E. Olive, "Differences in Performance and Preference of Trained Versus Untrained Listeners in Loudspeaker Tests: A Case Study," J. AES, Vol. 51, issue 9, pp. 806-825, September 2003. (download for free courtesy of Harman International).
[5} Sean E. Olive Sean E. Olive, "A Multiple Regression Model for Predicting Loudspeaker Preferences using Objective Measurements: Part 1 -Listening Test Results," presented at the 116th AES Convention, May 2004.
[6] Sean E. Olive "A Multiple Regression Model for Predicting Loudspeaker Preferences using Objective Measurements: Part 2 - Development of the Model", presented at the 117th AES Convention, October 2004.
[7] Floyd E. Toole " Loudspeakers and Room - A Scientific Review", J. Audio Eng. Soc., Vol. 54, No. 6, June 2006 (download for free here courtesy of Harman International)
[8] Todd Welti and Allan Devantier,"Low Frequency Optimization Using Multiple Subwoofers", J. Audio Eng. Soc., Vol. 54, No. 5, May 2006 (download for free, courtesy of Harman International).
Last edited: