This New Research Claims Headphones Have Been Tuned Wrong for Decades, but Critics Aren’t Buying It

The study tries to rewrite audio history with a massive headphone dataset, bold tuning claims, and zero patience for Harman’s legacy.

Stanford-based audio company PEQdB has released research challenging Harman International’s industry-standard headphone tuning curves.

The company claims its machine learning approach produces superior sound profiles. They used data from 266 headphone models and say this addresses methodological flaws in the current gold standard.

PEQdB Takes Aim at Industry Standard

When it comes to headphone tuning, Harman’s over-ear 2018 target curve and in-ear 2019 target curve have set the bar for years. But PEQdB thinks they’ve found a better way, as presented in their recent white paper.

This challenge enters an industry where Dr. Sean Olive has been publicly criticizing how reviewers misuse his work. Olive is head of acoustics research at Harman International and creator of the original curves. He warned that oversimplified rankings ignore crucial factors like listener segmentation and individual preferences.

PEQdB claims their approach addresses these concerns by testing across a much broader dataset. Harman’s original studies used between 10 and 249 listeners depending on the specific research.

Competing Methodologies: Different Equipment, Different Results

PEQdB and Harman use very different approaches. This shows they disagree about how headphones should be measured and tuned.

PEQdB criticizes Harman’s use of a “flat in-room” baseline. They argue it contains inherent flaws. The company points to Harman’s use of the GRAS 45CA measurement device. They call it “a flat-plate headphone measuring device with no human-like anatomical features beyond its anthropometric pinna.”

Instead, PEQdB uses a diffuse-field head-related transfer function (HRTF). This is based on the average eardrum response of 47 individuals and follows ISO 11904-1:2002 standards. This approach, they argue, better represents how humans actually hear sound.

However, PEQdB’s equipment choices have raised eyebrows. The industry is increasingly moving toward the Brüel & Kjær Type 5128 measurement system.

PEQdB continues using GRAS 43AG and 45BC test fixtures for over-ear measurements. They also use IEC 60318-4 couplers for in-ear measurements. The company explicitly avoids the B&K 5128 due to what they claim are “unnatural low-frequency resonances” in in-ear measurements.

The company tested nine variable parameters across three filters simultaneously. This compares to Harman’s approach of allowing listeners to adjust only bass and treble through two filters. PEQdB’s listeners made adjustments using “Inner Cell” by King Gizzard & the Lizard Wizard. The song was chosen for its “high bandwidth, spectral density, and dynamic range.”

Large Dataset vs. Recent Listener Studies

PEQdB’s massive dataset of 266 headphone models contrasts sharply with Harman’s smaller but more controlled studies. However, recent research suggests Harman’s approach may still connect with listeners.

A 2024 study involving 32 Harman employees found that 72% of participants favored sound profiles similar to the Harman IE 2019 target curve. The study compared five different target curves using the newer B&K 5128 equipment. It showed the Harman curve tied statistically with the SoundGuys curve for highest preference ratings.

The research identified two main listener segments:

• 72% preferred fuller bass and balanced highs (Harman-style)
• 28% favored less bass and brighter treble (typically older listeners who prefer “2dB less bass and 1-2dB more treble”)

PEQdB’s findings challenge this segmentation. Their data shows people prefer similar sound signatures whether using in-ear or over-ear headphones. This goes against Harman’s approach of creating separate curves for different headphone types.

Beyond questioning Harman’s segmentation approach, PEQdB also addresses common complaints about the curves themselves. Users often say Harman curves sound “thin in the bass,” “veiled in the highs,” and “shouty.” PEQdB says their curve solves these problems by using better testing methods and more data.

The company’s research also found no major difference between users who tested with their own music versus the default test track. This backs up earlier Harman research that found the type of music doesn’t change what sound people prefer.

What PEQdB Gets Wrong

While PEQdB presents an ambitious challenge to Harman’s long-standing target curves, it’s far from perfect. Critics have pointed out several flaws in both its research approach and the way it presents itself.

Here’s a closer look at where PEQdB might be missing the mark.

• Lack of scientific rigor
• Methodology limitations
• Questionable claims and tone
• Community skepticism
• The author factor

Lack of scientific rigor

One of the biggest concerns is that the PEQdB white paper isn’t peer-reviewed. It’s self-published and doesn’t follow the usual process of academic research. That doesn’t automatically mean the findings are wrong, but it does raise questions about how much we can trust the conclusions.

There’s also a noticeable lack of proper statistical analysis. There are no error bars, no confidence intervals, and very little explanation of how consistent the results are across different users or listening setups.

For something that claims to be “statistically optimal,” that’s a big gap.

In contrast, Harman’s research was published in journals and included detailed statistical modeling to support its conclusions.

Methodology limitations

The PEQdB tool allows anyone to take part in the testing, which sounds great at first. But without controlled environments, that flexibility becomes a weakness. People might be using the tool in noisy rooms, with poor gear, or without knowing how their headphones measure up to the versions in the PEQdB database.

There’s also no way to know if a user’s headphone is exactly like the one the tool expects.

Manufacturing differences, even between left and right channels, can affect the sound. So, if a company has poor quality control, PEQdB’s preset values might not match the headphones you actually own.

As for IEMs, the tool doesn’t account for different eartips, which are known to seriously change how bass and overall tuning are perceived. That’s a big oversight for something trying to nail down listener preferences.

Then there’s the way the tool works: users do A/B tests using short 20-second clips. That kind of short-term listening might not reflect what people prefer over longer sessions.

It’s easy to pick something that sounds good in the moment but gets fatiguing over time. That raises doubts about whether the results really show lasting preference or just first impressions.

Questionable claims and tone

The paper’s tone has also rubbed many readers the wrong way. Instead of building on existing research, it spends a lot of time criticizing Harman and presenting its own work as the new gold standard, without actually proving it.

Statements like “the most statistically optimal headphone target curves ever created” sound bold, but they don’t hold up when the methods haven’t been clearly validated or compared against the Harman curve in a controlled test.

Critics have called this kind of language arrogant and unscientific.

Even more frustrating for some is that the paper barely acknowledges the work done by Sean Olive and others. Harman’s research helped define how we measure and understand headphone sound, and brushing that off without a thorough comparison doesn’t sit well with many in the audio science community.

Community skepticism

The reception from the headphone community has been all over the place. Some users say PEQdB gives them decent results, but many others have found it disappointing. A common complaint is that the curve has too much ear gain around 3kHz, which can make music sound narrow or fatiguing. Others say it sounds too tame in the bass and treble.

There’s also a sense that the final curve isn’t all that different from Harman’s to begin with.

In fact, some users feel the study mostly confirms what we already knew, that many people prefer something close to Harman, with slight tweaks based on personal taste.

So while the tool is interesting and even helpful in some cases, it hasn’t convinced everyone that it’s a major breakthrough. Some think it’s just another way to end up at the same place.

The author factor

Finally, there’s the issue of the author, known online as Sharur. He’s a controversial figure in the audio community, with a reputation for stirring up drama and clashing with others in forums. That’s made it harder for some people to take the research seriously, even if the tool itself is useful.

It’s true that good ideas can come from difficult personalities. But when the presentation is filled with bold claims, dismissive language, and no real peer review, it becomes tough to separate the message from the messenger.

Some users even suspect the whole project is more about attention than science.