The crowd at the January 2010 AES PNW Meeting.
|Photo by Gary Louie|
The AES PNW Section and Art Institute of Seattle Student Section presented a January 2010 program featuring Dr. Floyd Toole and Dr. Sean Olive of Harman International, and a tour of the venue, Microsoft Studios in Redmond, WA. About 95 total persons attended (38 being AES members). Earlier that afternoon, the AIS students held their own discussions with the presenters at their school, in part because Dr. Olive also serves as AES Western Region VP. PNW Chair Steve Turnidge opened the evening meeting, conducted some AES business, and ran the traditional self introductions by the attendees.
First to present was Dr. Floyd Toole, who retired from his post as Harman VP for Acoustical Engineering 2 years ago (he is still a consultant to them). His mildly provocatively titled presentation was called "Upgrading the Audio Industry," but it focused on loudspeakers. He feels that both the pro and consumer speaker industries have largely been mired in unscientific misinformation to sell speakers for too long, and it's not necessary. His research, along with Dr. Olive's, shows that with the right measurements, we can describe real-world speaker performance scientifically. When combined with valid research on trained listener preferences, one can identify a good, neutral speaker, and users need not be mystified about how to find one. The audio world would be better off if everyone could identify, buy and use neutral, good sounding speakers, instead of having to throw the dice.
He has no problem with anything done to recordings for artistic creation of sounds, but he does believe that monitor and user listening speakers should be neutral, and not "musical instruments" themselves that can be liked for enhancing a recording to your preference. We can and should offer scientifically valid standards for speakers that ensure getting as good a unit as you can get for your money.
Sean Olive's following talk exploreded the validity of using trained listeners in double blind listening tests. A correlation between specific technical measurements and what people consistently identify as good sound means that the psychoacoustic link is established.
Dr. Toole went into some detail about the speaker tests needed. Sound power, directivity index, early and late reflections - all need to be accounted for. The results can be simplified into a user-friendly family of curves. His tests showed that simple advertising phrases such as, "20-20kHz ± 3dB" or a single response curve are fairly useless for identifying a decent real-world speaker. Equalization, while important in correcting lower frequency room effects, can't correct for poor off-axis response.
He noted that an automobile tire must legally have more useful information on it than a loudspeaker.
At Harman, the goal is trying to achieve good sound everywhere - the studio, home, car (a real challenge), and for desktop computers and iPods. Their research shows that psychoacoustic correlation can be made between the right technical measurements and double blind trained listener tests.
Dr. Olive described how listeners are trained to recognize and describe good sound, and how this absolutely does not bias them in their ability to recognize good sound in blind tests. Nuisance variables are controlled, such as knowledge of a brand name or price, as speakers are all behind a screen, and moved to the correct listening position by an automated speaker mover.
Objective measurements include a battery of some 70 tests in the anechoic chamber, covering vertical and horizontal responses in 10 degree increments. He developed methods to characterize many measurements and show graphs accounting for the quality of the direct sound and early and late reflections in a typical room.
Another recent study of his used similar double blind methodology to evaluate computerized room correction systems. Out of 5 systems tested, only 3 helped much, with 1 judged about the same as no correction, and 1 judge much worse.
Dr. Olive also spoke a little about doing similar automotive tests. Blind tests in a real car are impractical, so he described some ways they try to do meaningful tests, such as binaural room scanning.
Finally, he showed a bit of their listener trainer program (PC or Mac), where listeners are tested and trained to identify and describe characteristics of speaker sound.
Lastly, groups gathered for the Microsoft Studios tour from Rick Senechal and the MS Studios staff. Microsoft Studios is a full-service production facility, with three sound stages for live broadcast or stage production, as well as audio production studios. The facility also has three Starbucks coffee machines so the work can go on around the clock.
Reported by Gary Louie, PNW Section Secretary
Last modified 10/13/2011.