Put two headphones on the same measurement rig and you can get graphs that look almost identical in the treble: the same rough level, the same overall tilt. Put them on your head and one sounds open and airy, while the other makes every "s" spit at you. The graph didn't lie. It answered a different question from the one your ears are asking.

Treble differences on a graph come in two basic shapes. A broad lift raises a wide stretch of the spectrum, often an octave or more, by a few decibels. A narrow peak raises a small slice, sometimes only a few hundred hertz wide, by much more.

Averaged over the treble, the two can add a similar amount of extra energy. A gentle 2 dB lift from 4 to 12 kHz and a sharp 8 dB spike at 7 kHz can end up with similar numbers if you sum the region up as one figure. That's why one-word descriptions such as "bright" aren't very useful on their own.

Your ears treat the two very differently. A broad lift changes the tonal balance: the whole top end moves forward, cymbals shimmer, voices gain air, and the headphone sounds bright. A narrow peak picks out specific sounds and makes them jump out of the mix. One is a dimmer; the other is a spotlight.

Roughly 5 to 8 kHz is where much of the energy of "s", "sh" and "t" sounds sits, along with the attack of cymbals and hi-hats. Our hearing is also very sensitive across the upper midrange and lower treble, so a few decibels there go a long way.

Put a narrow peak in that band and sibilants hit harder than everything around them. On a first listen it can pass for detail, because certain sounds suddenly seem clearer. Over an hour it turns into fatigue: the same sounds keep jumping out, and you start reaching for the volume to make them stop.

There's a second effect. A narrow peak in a headphone usually comes from a resonance: a part of the driver, the cup or the air space in front of your ear that rings at a particular frequency. A resonance doesn't just add level. It stores energy and releases it slowly, so that frequency lingers slightly after the sound stops. The ear reads that lingering edge as hardness, which is why a peaky headphone can sound harsh even at modest volume.

A few habits make headphone graphs far more useful:

1. Look at the width, not just the height. A bump spanning an octave or more is tonal balance. A spike that rises and falls within a fraction of an octave is a peak.

2. Check the smoothing. Heavy smoothing, such as 1/3 octave, rounds narrow peaks into gentle hills. If a graph looks suspiciously smooth, find one with finer resolution.

3. Be careful above about 8 kHz. Treble measurements up there depend heavily on the rig and on how the headphone sits, and they shift with fit and pad wear. A single sharp spike at 10 kHz on one graph may be the rig, not the headphone. A peak at 6 or 7 kHz that shows up in several independent measurements is far more likely to be real.

4. Compare against a target, not against flat. Headphones aren't supposed to measure flat, so look at how far a peak sticks out from the target curve the reviewer uses.

For the background on how headphone frequency response graphs work and why they look the way they do, there's a detailed explainer here: https://theaudiostuff.com/glossary/frequency-response/

Reading about a peak is one thing; hearing it is quicker. Take a vocal track with plenty of "s" sounds and try two EQ changes, one at a time:

- A broad lift: a high shelf of +3 dB starting around 4 kHz.

- A narrow peak: +6 dB at 7 kHz with a narrow bandwidth (a Q of around 5).

The first sounds brighter. The second sounds wrong: sibilants spit, and the effect follows the voice around while everything else stays put. If you'd rather not set up a plugin, you can try both in the browser with this free EQ simulator: https://theaudiostuff.com/tools/eq-simulator/

Then try it in reverse. If your own headphones already sound sharp on sibilants, sweep a narrow cut of 3 to 6 dB through the 5 to 9 kHz range. When you land on the right frequency, the harshness drops away and the rest of the treble stays where it was. That's the clearest sign you were dealing with a peak rather than general brightness.

This is where it gets practical for anyone who records, streams or chats with a headset or a mic attached to their headphones.

When you monitor your own voice or play back a recording, you're hearing the microphone through your headphones. If those headphones have a peak in the sibilance band, every "s" in your recording sounds harsh, and the natural conclusion is that the microphone is sibilant. People then add de-essers, cut treble at the mic or replace it, and the recording ends up dull for everyone else while still sounding sharp to them.

Before blaming the mic, check the playback. Listen to the same recording on a different pair of headphones or on speakers. Look up a measurement of your headphones and see whether there's a narrow peak around 6 to 8 kHz. If there is, fix it with EQ on playback, not at the microphone, so your recordings stay natural for everyone listening on other gear.

Two headphones can carry the same amount of treble and still sound nothing alike, because width matters as much as level. Broad lifts change the tonal balance; narrow peaks pick out particular sounds and make them harsh. Read graphs for width, be skeptical above 8 kHz, and trust your ears once you know what to listen for. And if your mic sounds sibilant, check your headphones first.