Study Shows No Correlation Between Price and Sound Quality in Headphones

Over the last decade, the headphone market has seen a boom in sales and growth. This expansion in business has caused a lot money to be poured into research into headphone development for a better understanding of listener preferences. Much of the new research into headphone technology has led to some surprising conclusions, including one paper recently published by Jeroen Breebaart in the June 2017 issue of The Journal of the Acoustical Society of America. The findings of this paper can be mostly gleaned from its title: “No Correlation Between Headphone Frequency Response and Price.”

The research for that study measured the frequency response of 283 headphones over a wide range of pricing from $4 to over $5000 and found that the frequency response had a nearly insignificant correlation to the cost. In other words, just because you spend more money, doesn't mean you will get a pair of  headphones with more linear or better frequency response characteristics. In-ear, supra-aural, and circumaural headphones were the headphone types that were examined, and the study also showed that there wasn’t even much of a correlation of frequency response to headphone type.

One of the reasons why this is surprising is because frequency response was found to be the major determinant of sound quality in headphones, based on previous research by Sean Olive and Todd Welti. Some might wonder if other attributes of sound reproduction play an equally important role as frequency response, but this does not appear to be the case. Nonlinear distortion like harmonic distortion can crop up when a headphone is driven hard enough, but most headphones seem to have enough headroom for that not to be an issue. We asked Dr. Sean Olive about how much nonlinear distortions would be a factor that could separate the sound quality of different headphones.

He replied, “Based on our research and others, I believe frequency response is the dominant factor that determines sound quality. We've done some experiments where we equalized the headphones to have the same frequency response so that listeners could focus on distortion differences. The headphones sounded sufficiently the same that listeners could not say they preferred one to the other except in 1 or 2 cases where the distortion was really high.” Time-related properties such as group-delay would also seem not to be an issue in headphones, since most headphones are minimum phase up to high-frequencies where it is not audible, and any changes in the phase response would necessarily cause a change in the frequency response.

What Olive and Welti found was headphones that were perceived  to have the most neutral frequency response were preferred by trained listeners. In other words, much like research conducted for loudspeakers, listeners gravitated toward accuracy. In view of Breebaart’s finding’s, this then begs the question of what is the sound quality advantage of purchasing expensive headphones?

Digging a bit deeper into Breebaart’s paper, there were some minor differences in price and headphone type. For example, the more expensive headphones did not seem to have quite as much variance in low-frequency response. This might be due to manufacturers targeting a more neutral response, but it may also be due to the more expensive headphone having a better seal around the test ears and therefore a more consistent bass response. Also, in-ear headphones had a greater levels of low-frequencies, although that might be due more to testing conditions than what would be experienced in normal listening conditions. A dummy head was used for all testing that might have had a greater acoustic seal than normal human in-ear use.

One interesting aspect that Breebaart uncovered was how closely the average response of all the headphones followed that of an accurate loudspeaker’s in-room response, the ‘neutral’ response. According to the research of Sean Olive, Todd Welti, and Elisabeth McMullin, the response of an accurate loudspeaker in-room was found to be very close to the preferred response by listeners. However, when the average of all measured headphones in Breebaart’s survey was compared to the target response curve devised by Olive, Welti, and McMullin, the average did have more energy in the 50 Hz to 2 kHz region than the target curve. The average curve also had substantially less deep bass below 30 Hz than the target curve. It should be stressed here that the averaged curve deals with a lot of different headphones, and their response varied greatly. You can’t pick a set of headphones at random and expect its response to resemble the averaged response curve from this measurement set.

Another feature that emerged from Breebaart’s study is that the average response of circumaural and in-ear headphones more closely matched the ‘accurate loudspeaker in-room’ target curve than the average response of the supra-aural headphones.

The greater meaning of Breebaart’s study, and building upon Olive and Welti’s research, is that the price of a set of headphones is not a very significant indicator of its sound quality, assuming that the deviation from a ‘neutral’ response is a valid descriptor for the perceived quality. While it is true, as mentioned before, that the more expensive headphones exhibited somewhat less variance in their responses than the inexpensive headphones, the difference between the two was not tremendous. And, as the author states among his conclusions, “there are plenty of relatively cheap models that match the assumed target function, as well as very expensive ones that deviate significantly from an assumed ideal response.”

Furthermore, a capable enough headphone can reasonably simulate the qualitative sound of any other headphone with equalization. Olive, Welti, and McMullin did just this in their study “A Virtual Headphone Listening Test Methodology” where they used a plethora of filters to make their Sennheiser HD 518 headphones sound like other headphone models. In light of this, it looks to be possible to have a inexpensive headphone mimic the sound quality of an expensive headphone, so long as the frequency band range and dynamic range are available. In fact, this is the goal of the TB Morphit, a VST plug-in that can shape the sound character of a wide variety of headphones to imitate that of another model, or aim for another response such as more neutral responses. One of the reasons for Breebaart’s study is to determine if there was a case for TB Morphit, ie. software that could use one pair of headphones to emulate others, or, as Jeroen told me, “One of the main reasons for looking into headphone frequency responses was actually to see if there would be a case for headphone correction tools, and whether there is a specific price category for which that would be useful.”

Reportedly TB Morphit works very well in recreating the sound of one headphone in another headphone. There might be some small differences, however, such as manufacturing differences in the measured model or user’s headphones. Also, if the headphone was seated differently on the test dummy’s head than the user’s head, that could cause a slight difference. Aside from these differences, which are not likely to be severe, TB Morphit should recreate the sound character of any headset on any other headset in its list of headphone profiles. From a sound quality perspective, with TB Morphit, once you have purchased any headphone of their list, you can then listen to any other headphone on that list of profiles. Buying one headset with TB Morphit allows you to experience the sound of all 143 other headphones profiled, along with a handful of generic profiles including the ‘neutral’ curve. More headphone profiles are expected to be added in the future.

Why Buy Expensive Headphones?

Confronted with this fact, the only reason to seemingly purchase a high-priced headphone is comfort, build quality, or luxury amenities. It would appear that its ‘sound’ is being removed from the equation, since it can easily be replicated by a less expensive headphone. The question then is what impact will this research have on headphone consumers? What impact will it have on headphone manufacturers? How will this line of research change the headphone market when quality is merely a matter of frequency response, and frequency response is so malleable that the sound of one headphone can effectively be transformed into another? We are hoping to see this research acknowledged industry wide and is used to push the state of the art forward.