Our Headphone Measurement Database is Finally Here

After years of behind-the-scenes development dating back to 2021, GraphWorks is officially live to the public.

For years, audiophiles have asked why The Headphone Show never spun up our own squig.link database. The reality is that, as we've learned more and more, we've grown increasingly frustrated with how headphone measurements are typically presented across the internet, whether on squig.link, Audio Science Review, Reddit, Discord, YouTube, or anywhere else they're posted. Audiophiles have spent years taking graphs they barely understand at face value, when they really shouldn't. 

GraphWorks was built from the ground up to fix this habitual trap that hundreds of creators and consumers fall into every day.

By our estimation, the primary issue plaguing headphone analysis is the single-line measurement paradigm. Typical graphs plot a single frequency response line (or a single left and right trace) against a single static target curve. When viewers see a curve hug the target, they automatically assume the headphone sounds good, and when they see it deviate, they assume it sounds terrible.

That conclusion is rarely justified—especially not by the listener preference research that produced the target(s) most people use—because a single line measurement only represents one seating under one specific set of conditions, and a single line target only represents the average, not the spread of possibly preferred curves.

With headphones in the real world, positional variation can alter the response considerably more than a single line measurement conveys, and testing the same headphone across different measurement fixtures or actual human heads reveals an even wider distribution of possible acoustic results.

In actual listening tests, subjects rarely all prefer exactly the same thing, and even an individual subject usually has multiple sound profiles they'd call enjoyable.

A single line for both measurement and/or target creates an illusion of certainty that the truth of these devices and their interaction with human ears & brains simply do not support, which means most measurements out there are unintentionally acting as a catalyst for false confidence.

GraphWorks is designed to replace that false certainty with a more realistic approach to headphone data. The platform displays positional spreads and variation envelopes in a rig-specific Diffuse Field compensated environment that ensures that the viewer isn't getting any undue confidence from the visualization, especially not false confidence.

The core database is entirely free for anyone looking to compare headphones, while a $5 monthly supporter tier unlocks deeper visualization features, on-head human measurement datasets collected via in-ear microphones, and one more incredibly important feature: Our new EQ. 

To translate our data-borne insights into what ultimately matters most—improved sound quality—GraphWorks integrates directly with our new GraphWorks EQ for macOS and Windows.

Rather than manually exporting text profiles into third-party software, GraphWorks EQ links directly to the measurement database, letting you affect the headphone's response in real-time and visualize any changes you're making effortlessly.

GraphWorks EQ also features a built-in tone generator, which is an absolute necessity for using measurement data for EQ, because the data will almost never perfectly align with what the end-user perceived. An integrated sweep tool for a sine wave or band-limited pink noise allows listeners to identify  peaks and resonances directly at their own eardrum, making true perceptual equalization accessible and intuitive.

Alongside the software rollout, GraphWorks introduces our updated 2026 Preference Bounds for headphones and IEMs, the latter of which is in Beta. 

Moving away from all of the assumptions that underpinned our original Preference Bounds (interrelatability of headphones and IEMs, equivalence between Diffuse Field and "Harman In-Room Flat", simplified shelf-filter approximations), the new headphone Preference Bounds are built directly on the distribution of curves that listeners actually preferred in Harman's foundational 2018-2019 listening tests, while the new IEM Preference Bounds are built on the totality of Sean Olive's studies on listener preference in IEMs.

The Bottom Line

Sound quality is not a monolithic, one-size-fits-all line; listener preference exists, and has always existed, across a defensible range of responses. We absolutely rejects the idea that measurements should be discarded in favor of blind faith in "trust me bro" impressions around the internet, but instead of just rolling with the waves of "good enough" measurements that this community is currently misusing, we're providing a more sophisticated toolkit that reflects how headphones actually interact with human hearing, and hopefully puts a damper on the propagation of false confidence based on misused data.