PNW Sept 2014 - left/ right, Dave Tosti-Lane (PNW Vice chair), Dr. Sean Olive (Harman), Steve Malott (PNW Chair)
PNW Section started its meeting season with a presentation by Dr. Sean Olive of Harman International and President of the AES. About 57 (28 AES members) attended his presentation on research into headphone quality, held at Shoreline Community College in Shoreline, WA near Seattle.
Chris Deckard won a special election to fulfill the PNW Section Committee term of Scott Mehrens, who moved to California.
Sean described several studies and papers and 2 years worth of research either presented or to be presented, on headphone sound quality. Headphone sales are now booming, but there is little research or standards on quality and listener preference, not unlike loudspeakers until Sean's recent research at Harman. There are a few measurement standards, but no adherence by manufacturers of consumer headphones, and seemingly driven mostly by marketing.
The first study investigated whether trained listeners agree on what makes headphones sound good. They tested several models in many prices and several styles (open, closed etc.). Listeners didn't know which ones they were fitted with, but could eventually identify some by fit characteristics such as clamping force. Trained listeners (by a Harman training course) identified perceived spectral balance despite measured (ITU pinna by GRAS) response. Fit and seal affected each trial a lot. Listeners gave higher ratings to phones with smoother spectral balance.
The next study investigated listener preference to headphones equalized to certain targets. Known targets included equalizing at the eardrum like a speaker in a diffuse room; or a free-field room; or some variations. Harman created a new target response by measuring ear response to a reference 7 speaker system in a reference room. Most listeners preferred the new Harman curve over diffuse, free field or un-EQed.
Another study had listeners change bass and treble to preference for measured flat speakers and for headphones. Trained listeners boosted bass and treble less than untrained. People prefer some bass boost (they likely want room gain, like the original recordings) and a little treble cut.
The next study investigated the use of virtual (simulated) headphones. One reference headphone is equalized to simulate other models. This makes comparison tests much easier, but doesn't include nonlinear distortions or fit comparisons. 6 models were tried. This works well up to 10kHz or so, before physical differences affect the simulation. Listeners gave higher ratings to real headphones than virtual.
Sean didn't elaborate on a study of college student preference, but another study responded to the popular press saying young people preferred the sound of lower fidelity MP3s. This was apparently based on one unscientific article. Sean brought in high school and college students to Harman's lab; the majority chose CD quality over MP3, listening to speakers. Sean decided he could also test if they liked accurate speakers. Using the Harman speaker randomizer, accurate speakers were preferred, although experienced listeners rated all speakers with lower scores compared to the scores from less experienced listeners. The Harman randomizer moves one of several speakers behind a screen into the listener sweet spot, allowing double blind testing.
Another study examined cultural preference for headphone sound. Subjects in four countries tried virtual headphones using a custom Harman iPad app. The preference was for the accuracy favoring the Harman target curve.
In response to marketing, Sean felt that if designed to sound good to trained listeners, a headphone will also sound good to the general population.
Q and A continued with remarks on testing standards, artificial pinnae vs. flat plates, 3-D printed pinna, and hearing aid problems. Following a refreshment break and prize drawing, the remaining time was spent on more Q and A and discussion.
