With hundreds of new headphones coming out last year, S+V got so wrapped up in product reviews that we never got around to covering what I think is probably 2012's most important audio story: a recent research project that should augment the audio world’s spotty understanding of how headphones should be voiced.
"The Relationship between Perception and Measurement of Headphone Sound Quality," a paper by Sean Olive and Todd Welti of Harman International delivered at last October's Audio Engineering Society convention in San Francisco, should be as important in the headphone world as Einstein’s "On the Electrodynamics of Moving Bodies" was in the physics world. It won’t be, because there are loads of companies that see headphone development as little more than an exercise in marketing and industrial design. But at least the paper will give smart, motivated engineers and manufacturers needed guidance in the development of headphones that the average listener will like.
TRBPMHSQ seeks to find out what types of headphone tonal balances and voicings are generally the most pleasing. It starts with what may be the most careful and stringent listening tests ever done on headphones. Trained listeners evaluated six different over-ear headphones in a blind test. The headphones were placed on listeners’ heads by the test administrators, and plastic handles were attached to the earpieces so the listeners could adjust the headphone position for best sound and comfort without getting any clue as to the headphone’s identity. (Of course, the listeners could still get some idea of each headphone’s size and configuration because they could feel the headphones’ ear pads on their cheeks, a flaw Olive and Welti acknowledge.)
The headphones’ frequency responses were then measured on a G.R.A.S. Model 43AG ear/cheek simulator (the same one we use here at S+V). These frequency responses were compared with the results of the subjective tests to see what types of voicings and tonal balances gave the most pleasing results. The headphones tested were the $299 AKG K550, the $299 AKG K701, the $999 Audeze LCD2, the $299 Beats Studio Limited Edition, the $299 Bose QC-15, and the $199 V-Moda Crossfade LP. Olive and Welti found a strong correlation between the measured responses and the subjective test results. Their primary finding: “The most preferred headphones generally had the smoothest and flattest measured amplitude response when measured on G.R.A.S. 43AG ear simulator.”
In other words, headphones that had a fairly flat measured response through most of the audio range—and, most important, an even balance of bass, midrange, and treble—scored the best. Headphones that exhibited large dips and peaks in the response, or those that had the strong response peak around 3 kHz (recommended in the IEC diffused-field calibration, today’s most commonly used headphone voicing curve), didn’t score as well.
Olive and Welti didn’t reveal the scores of the individual headphones, but having measured all of the headphones myself (except in the case of the LCD2, but I've measured the similar LCD3), it looks to me like the clearly preferred headphone was the Audeze LCD2, followed by the AKG K701 and (surprise!) the Bose QC-15. (So much for the assumption among some enthusiasts that noise-cancelling headphones suck.) Headphones with pumped-up bass and large midrange dips (the Beats Studio and V-Moda Crossfade LP, apparently) didn’t do so well.
It's important to note that this is just one test with six very different headphones and nine trained listeners, in a quiet environment using two pop tunes and a classical piece as test material. It doesn’t show us, for example, what kind of headphone frequency response would most please an 18-year-old hip-hop fan riding on the subway. And of course, it's no big surprise when a $999 headphone outperforms a bunch of $299 headphones.
Another paper delivered at the same AES conference, "Identification and Evaluation of Target Curves for Headphones," contradicts Olive and Welti's findings to some extent. "Flat responses are not preferred, while highest ratings of quality are seen with equalization curves adjusted by expert listeners," the paper states. But the methodology used in the cited study—equalizing a set of headphones for flat response, then comparing the flat response against two established equalization approaches and EQ curves created independently by three experts—seems to me more complex and prone to error than the methods used by Olive and Welti.
Olive and Welti acknowledge that TRBPMHSQ is just a start, and suggest future tests involving trained and untrained listeners, more types of headphones, etc.
Meanwhile, here's one small confirmation of their findings: The headphone that got the biggest raves last year from S+V’s testing panel—the PSB M4U 2, our 2012 Product of the Year—has one of the flattest frequency responses we’ve measured in a headphone.
Only time will tell how much impact TRBPMHSQ will have on headphone design. But maybe it’ll at least help put to rest the idea that recording or producing a few hit tunes makes you an expert headphone tuner. If the world has just a few fewer headphones with unrealistically boosted bass response, the efforts of Olive and Welti will have been worth their while.
Brent Butterworth and Geoff Morrison combine their years of gear testing and knowledge in one überblog of irreverence and techiness.
Copyright © 2013 Bonnier Corp. All rights reserved. Reproduction in whole or in part without permission is prohibited.