The Harman preference score predicts how much listeners would like a headphone’s sound from one measurement: its frequency response. It comes from blind listening research at Harman International, led by Sean Olive, in which hundreds of listeners rated headphones they couldn’t see. The ratings were modeled against how far each headphone’s measured response strayed from the target curve those listeners preferred. Every one of the 666 headphones on this site is ranked by this score.
A higher score means the headphone’s tonal balance sits closer to what most people preferred in those tests. It is not a percentage and not a review rating; it is a predicted preference on a scale where the best headphones land around 90 and poorly tuned ones fall well below 50.
What Goes Into the Score?
The model compares a headphone’s measured frequency response with the Harman target, an in-room-like tuning with a gentle bass shelf and a smooth treble roll-off. It then uses two numbers from the difference between them:
- How uneven the error is. This is the deviation from target, the standard deviation of the difference curve in decibels. Peaks, dips and ripples raise it.
- How tilted the error is. A headphone that is uniformly too bright or too warm across the band gets a tilt penalty, shown on each model page as Tonal Tilt.
Smaller deviation and less tilt mean a higher score. Over-ear and in-ear headphones each have their own formula and their own target, so compare scores within a type rather than across them.
Which Frequencies Count?
The over-ear model scores roughly 50 Hz to 10 kHz, and the in-ear model roughly 20 Hz to 10 kHz. Above about 10 kHz, measurements depend heavily on the test fixture and each listener’s ears, so the research leaves that region out. Below 50 Hz, over-ear results swing with seal and head size.
That means the score can miss how deep a headphone’s bass really reaches. Each model page adds a Sub-bass Extension spec for exactly that reason, and the deep sub-bass headphones page collects the ones that stay close to target down to 20 Hz.
Where Do the Measurements Come From?
Most scores on this site come from AutoEQ, which applies the Harman model to measurements published by oratory1990, crinacle and other reviewers. Models measured only by hobbyists on squig.link use the median of every measurer’s curve, calibrated so the scores line up with AutoEQ’s. Each model page names its source in the Harman Preference Score row.
What the Score Does Not Tell You
The score knows nothing about comfort, fit, build quality, noise cancelling, Bluetooth features or how a headphone handles very loud playback. In-ear results also depend on getting a good seal with the right ear tip. Treat the score as the best single guide to tonal balance, then check the specs and the pros and cons for everything else.
A lower-scoring headphone can often be improved a lot with EQ. Most model pages include an EQ preset that brings the response closer to the target.
Related Speaker Guides
- Deviation From Target Explained
- Spinorama Preference Score Explained
- Estimated Speaker Ratings Explained
- Bass Extension Explained: -3dB vs -10dB Meaning
- Speaker EQ Explained: How Many Bands Do You Need?
Frequently Asked Questions
Is a higher Harman score always better? For tonal balance, generally yes: it predicts more listeners will prefer it. Small gaps of a few points are hard to hear, and personal taste still matters; the headphone buyer’s guide covers warm, bright and V-shaped alternatives.
Why do some great-sounding headphones score low? Some are tuned deliberately away from the target, for example with extra bass, and some reviewers’ favorites suit particular tastes. The score measures closeness to the average preference, not quality in every sense.
Can I compare an over-ear score with an in-ear score? Not directly. The two types use different targets and different formulas, so compare within the over-ear or in-ear rankings.