What is sibilance, and how do you fix harsh S sounds?

Where the hiss in an S comes from

A vowel starts as a buzz from the vocal folds, shaped by the throat and mouth. An "s" has no buzz at all. The tongue forms a narrow channel just behind the upper teeth, air is pushed through it, and the stream turns turbulent as it rushes past the teeth. Turbulence produces noise spread over a wide band with no pitch of its own, and in an "s" most of that energy lands in the treble, roughly between 5 and 10 kHz depending on the voice.

The related sounds are variations on it. A "z" adds the buzz back. A "sh" is made further back with rounded lips, which lengthens the space in front of the channel and moves its energy lower. "Ch" and "ts" release a stopped "t" straight into the hiss. "S" and "sh" are also much louder than an "f" or a "th", so they're the consonants that poke out first when the treble is too strong.

At the right level they make words crisp. They turn into sibilance when their band plays too loud against the rest of the voice: each one jumps out of the vocal line with a spit or a sting, and over an album the effect wears you out.

Why two listeners hear it differently

The ear canal is a short tube closed by the eardrum, and it resonates, lifting sound between about 2 and 5 kHz on its way in. Every headphone is tuned with that lift in mind. Above it, the response at your eardrum depends on the shape of your outer ear, the length of your canal and exactly where the pad or tip sits, and all three vary from one person to the next.

So the same headphone can put a treble peak at a slightly different frequency, and a different height, in two people's ears. Measured graphs are least reliable in the same region, for reasons the frequency response explainer goes into, so a small treble spike on a chart is weak evidence either way.

Reviews can disagree about sibilance for these reasons and both be accurate. The Sivga SV021 Pro review notes that other reviewers called its highs fairly exaggerated, while its own listening found the low and mid treble restrained and only the very top slightly hot, audible when a trumpet hits or a heavy cymbal crash lands. Sivga later sent a revised driver, reworked to address that sibilance, which the review heard as a little better in the very high frequencies.

Is it the recording or your gear?

Start with the music, because a lot of sibilance is recorded in. Vocals are often sung close to a bright condenser microphone, compressed, then brightened again in mastering. Engineers fight it at each stage: with a de-esser, which ducks the level each time a sibilant goes past, and with the microphone itself. The Austrian Audio OC16 review describes a response with a deliberate dip between 6 and 7 kHz, "exactly where S sounds get sharp". When that work falls short, the sibilance is on the recording and every headphone will play it.

So play five or six recordings you know, from different producers. If two of them spit and the rest don't, those recordings are the cause. A headphone with a treble lift pushes the already hot ones over the line first. The HIFIMAN Edition XS review describes a large boost in the high frequencies, and highs that sound plasticky and sibilant "in some songs".

If every track spits on one headphone and none does on another, look at the headphone and its fit. On in-ears, the tip's size, material and insertion depth all move the treble, as the IEM explainer describes. On over-ears, shift the cups a little on your head and listen again.

Electronics are the least likely cause, with one well-understood exception. If your source has a high output impedance, it can raise or lower a multi-armature IEM's treble, because that kind of IEM's impedance swings across the band. The impedance explainer sets out the check.

What "hot" and "splashy" mean in a review

Reviewers use "sibilant" first for consonants and often for the same band on instruments, as the SV021 Pro review does with trumpets and cymbal crashes. "Hot", "sharp", "spitty" and "splashy" point at the same region. "Bright" is broader: a lift across the whole treble, which may or may not bring sibilance with it.

Read praise for a sibilance-free top end closely too. Sometimes it comes from a smooth treble, and sometimes from one that sits a little back. The HIFIMAN Sundara review ties its lack of sibilance to highs a little below the target curve, the same balance that tilts it slightly warm. If treble bothers you, weigh a review's notes on sibilance above its praise for detail. For in-ears, where the tip decides so much of the treble, the best IEMs guide ranks the models reviewed here.

Cutting it with EQ

A treble shelf is the obvious move, and it tends to disappoint. A shelf lowers everything above its corner, cymbal shimmer and air included, while the narrow peak causing the trouble stays the loudest thing left up there. A narrow cut aimed at the peak takes out the sting and leaves the rest alone.

With a parametric equalizer the peak is easy to find. Play a sibilant track, sweep a narrow boost of a few decibels slowly between 4 and 10 kHz until the S sounds are at their worst, then turn that boost into a cut of a similar size. Use the smallest cut that removes the sting.

The site's EQ simulator has ten fixed bands, each about an octave wide, so it's a quick way to hear whether a small cut on the 8 kHz slider, or on the 4 kHz one, helps a track before you commit to anything. It exports the curve for Equalizer APO, AutoEQ, Wavelet and Poweramp, and a parametric equalizer such as Equalizer APO lets you narrow the filter onto the exact frequency.