Dr. Sean Olive is a Senior Fellow, Acoustic Research, at Harman International. He has extensive expertise related to perception and measurement of sound quality, and oversees Harman’s corporate R&D acoustics group. His objective and subjective critical listening research covers a depth and breadth of insights few can rival, with the result that many listeners enjoy the fruits of his labors.
In this two-part interview, we scratch the surface of some of this research, including insight into the Harman Curve for headphones, which is a frequency response curve that has been determined by Harman to be pleasing to a majority of listeners.
Russell Welton: What were some of the challenges to get AKG (a Harman brand) to implement the Harman Curve in some of their headphone models, including the AKG K371?
Dr. Sean Olive: Our research into the perception and measurement of headphone sound quality covered a seven-year span (from) 2012 to 2019. The AKG K371 was released in July 2018 and was the first AKG Professional headphone to implement the Harman target curve. Part of the challenge you mentioned was related to AKG going through some business and structural changes, which eventually led to their Vienna headquarters being transferred to California, where we are located. The new AKG management team was very familiar with our research and were anxious to transfer it into new headphone models: the AKG K371 and K361. The K371 received almost instant critical acclaim for its sound quality and helped to legitimize the Harman target curve as a reference benchmark.
RW: Can you tell us a little about the studies which correlate personality types with frequency response preferences? Does this influence your design and marketing strategies?
SO: Our research identified three segments of listeners based on their preferred headphone sound profiles, as well as the underlying demographic factors (age, gender, and listening experience) associated with each segment. The largest segment (64 percent) prefers the Harman Target Curve and includes all age groups, [levels of] listening experience, and genders. The second largest segment (21 percent) preferred the Harman target with 2 – 3 dB less bass and includes a disproportionate percentage of females and older listeners. The smallest segment (15 percent) preferred the 4 – 6 dB more bass than the target and consisted of mostly younger males.
Our JBL-branded headphones tend to have 2 dB extra bass than the Harman target curve because they are aimed at a younger demographic identified in one of the segments. The Mark Levinson No. 5909 headphones allow you to choose among three different bass levels so that it can satisfy the tastes of all three segments.
The Harman curve. Courtesy of Jazz Times magazine.
Our research specifically address the role of personality in preferred sound profile, but other studies have shown that psychoticism, gender, and extraversion are all positively correlated to a preference for enhanced bass. Younger males tend to make up that segment. Think about how many times you have encountered a car driving by with booming bass that rattles the windows, and there was a female driving it. Never? Other studies have shown musical preferences tend to be associated with your five-factor personality traits, and I wouldn’t be surprised if there is a connection with what type of sound you prefer.
RW: How significant is the new research in mapping inner ear canals for simulating listening tests as conducted by B&K using their Type 5128 high-frequency head and torso simulator (HATS)? [The type 5128 is an artificial “dummy head” and torso that is used to simulate the acoustic properties of an actual person – Ed.]
SO: The main advantage of the new B&K ear simulator is it more accurately reproduces the acoustic impedance seen by headphones below 100 Hz and above 8 – 10 kHz. In theory, this means that headphones measured on it will [have] a more accurate representation of how they measure on humans. While this is true as far as ear canals go, there are other factors that influence how a headphone measures on humans.
We recently measured nine models of headphones on different test fixtures and humans and found significant discrepancies at low and high frequencies. Since we measured the headphones at the blocked ear canal, the measurement differences were related to how the headphones couple and seal around the ear and face. This is particularly true for closed-back headphones that rely on a good seal for their bass.
Mannequin test fixtures like HATS and [the G.R.A.S.] KEMAR tend to be leakier than humans and underestimate the bass produced by the headphones [as] measured on humans. Test fixtures that use a flat plate with a pinna (e.g., the GRAS 45CA test fixture) tend to provide a better seal on the headphone than humans and overestimate how much bass humans will hear. So, there is still more work to be done on test fixtures. In the meantime, my advice for consumers is to always check the fit and seal of a closed-back headphone and listen to the bass balance before purchasing it.
Although the Harman target curve research was based on using the older IEC 60318-7 ear simulator, it doesn’t invalidate or change the target, since it was based on listening tests – not measurements. It only means that a headphone tuned to the Harman target curve will measure differently on a B&K 5128, particularly at low and higher frequencies. We have a Harman target curve defined for the B&K 5128 and may eventually publish it.
RW: What do you see as the future for the Harman Curve as the benchmark reference frequency response for headphones?
SO: I believe the Harman Curve has become popular and adopted as a benchmark reference because there is a significant body of published scientific research to back it up. But we don’t claim that it is the last word in headphone sound quality. The segmentation research we did suggests there are good reasons to slightly deviate [from] the target to satisfy different tastes based on age, hearing or listening experiences. I’m quite certain that other researchers – maybe even us – will propose alternative targets if new research supports it.
RW: What is pinna gain, and how does it affect different listeners? [The pinna is the external part of the ear – Ed.]
SO: Pinna gain is probably not the best term since the transfer function from a sound source to the eardrum is influenced by many anatomical parts besides the pinna. The well-known figure below summarizes the contributions in gain (dB) from the head, neck, torso, and external ear including the pinna. Together, they provide up to 17 dB gain at 2.7 kHz measured at the eardrum for a sound source located at 45 degrees to your left or right.
The largest gain occurs from the resonance of the ear canal at 2.7 kHz, followed by the concha (the central part of the ear) at 6 kHz. The black curve is often referred to as the head-related-transfer-function (HRTF) which defines the transfer function of a sound source to the eardrum. An entire set of HRTFs captures sound sources at different distances measured over a sphere; a binaural renderer can then simulate a 3-D sound field.
Everyone has a different set of HRTFs due to differences in the shape and size of their head, torsos and ears. Research has shown that the most accurate localization and timbre of binaural rendering happens when it’s personalized using the listener’s own HRTFs.
RW: What are your feelings about the inclusion of a crossover in IEM (in-ear monitor) and on-ear headphone designs?
SO: Crossovers are necessary if you are doing a multi-way design where two or more transducers are used to cover different parts of the audio band. This typically happens in IEM designs where, for example, a dynamic driver may be used for bass, and a balanced armature is used at higher frequencies. You need a crossover to optimize the frequency response of the system, compensate for differences in the [drivers’] sensitivities, and make sure the [drivers] are not operating in a region where they may produce distortion.
RW: What does your research reveal about the pros and cons of using planar magnetic design in IEMs?
SO: I don’t have much experience with planar magnetic designs in in-ear monitors. But if they meet all the necessary technical parameters to achieve good sound, then it comes down to tradeoffs in sensitivity, SPL output, weight, manufacturability, and cost. There are many examples of excellent planar magnetic headphones in the market, but they tend to be larger, heavier designs that have lower sensitivities and higher associated costs. For a high-end headphone intended for home use, that isn’t necessarily a big issue, but for a mass consumer product that intended to be small, portable, with good battery life, it can be a deciding factor.
In Part Two of our interview, Dr Olive discusses the further technological improvements we can look forward to for enjoying our recorded music, and offers advice for aspiring audio engineers and audiophiles.
Header image courtesy of Dr. Sean Olive.
