Split-band vs single-band de-essing

If you've compared two de-essers and one kept the air while the other dulled the whole top end, you've heard the difference between split-band and single-band de-essing. It's the most important architectural choice a de-esser makes, and it explains most of why some plugins sound transparent and others don't.

Key Takeaways

• A single-band (wideband) de-esser ducks one broad slice of the top end when it fires. Simple, but it dims more than the ess.
• A split-band de-esser divides the sibilant region into narrow bands and reduces only the one that's flaring, so the surrounding air survives.
• Narrower and more bands generally means more preserved air, up to a point.
• Sibilance uses 8 targeted bands across roughly 3.6–11 kHz.

What single-band de-essing does

A single-band de-esser watches the high frequencies and, when an ess crosses the threshold, pulls down one broad band, or in the simplest designs the whole signal. It's easy to build and easy to use, and for a lot of vocals it's fine. The catch is collateral: when it clamps to tame a 7 kHz ess, it also pulls down everything else sharing that band, including the vowel's air a couple of kilohertz away. Push it hard and the vocal goes smooth and a little lifeless at the same time.

What split-band de-essing does

A split-band de-esser divides the sibilant region into several narrow bands and reduces only the band where the ess is actually flaring. If the ess lives at 7 kHz, the 7 kHz band moves and the bands around it hold still. The air at 10 kHz stays put; the presence at 5 kHz stays put. You get the reduction where you need it and keep the frequencies you don't.

That's the whole reason split-band sounds more transparent: it's surgical where single-band is broad. The more precisely a de-esser can isolate the offending slice, the less of the vocal it disturbs to get there. We go into why that transparency matters in the complete guide to de-essing.

Narrower isn't infinitely better

More bands and narrower cuts preserve more air, but there's a limit. Bands that are too narrow can ring or sound unnatural, and past a certain resolution you stop hearing a benefit. The goal isn't maximum bands for their own sake; it's enough resolution to separate the ess from the air around it, cleanly. In practice a handful of well-placed bands across the sibilant range does the job.

Sibilance splits into 8 targeted bands across roughly 3.6 to 11 kHz, with a distributed multi-notch, so only the sibilant bands move and the rest of the top end is left alone. On our test material that keeps the non-sibilant highs within about ±0.04 dB of untouched, and we publish the measurement rather than asking you to take "transparent" on faith.

Which one do you have?

• If your de-esser has a single frequency and threshold and dulls the vowel when you push it, it's single-band. It'll get you through most mixes; just use a light hand.
• If it splits the high end and only touches the offending region, it's split-band, and it'll let you de-ess harder before the air suffers.

This is also part of why a phoneme-aware split-band tool beats a manual dynamic EQ band: the dynamic EQ gives you one band to aim, while a good de-esser is already splitting the region and following the ess for you.

The short version

Single-band de-essing ducks a broad slice and takes some air with it; split-band de-essing isolates the ess and leaves the rest of the top end alone. For a vocal you want smooth and open, split-band is the architecture that gets you there. If you want to hear it on your own take, Sibilance has a free tier and splits into 8 bands so the air survives the de-ess.

Sibilance is the single-authority de-esser this comes from — a real free tier, and a 7-day Pro trial with no card.

Related reading