Smarter Sibilance Handling For Bright Vocals And Tough Dialogue

Sibilance is a problem faced by those working in both music and post-production, and we've had tools to deal with it for years. However, sometimes those tools can cause as many problems as they fix.

A typical de-esser works as a frequency-aware compressor. You set a threshold and choose a band; it reduces the level when the energy exceeds that point. This usually works for simple material, but in real sessions, more demanding situations come up fast. The standard method then starts to show its limits, and you end up sacrificing either the fix or the sound.

For example, bright female vocals with lots of high-end, or dialogue recorded with a hyper-cardioid mic in a reflective room. Sometimes a singer's sibilance is at 7 kHz in one phrase and 9 kHz in another. In these cases, a static de-esser makes you choose between letting harshness through or losing the presence and clarity the vocal needs.

Why The Broadband Approach Causes More Problems Than It Solves

The problem isn't with detection, but with what happens next. When a standard de-esser lowers a 2 kHz-wide band, everything in that range drops together. This means the "s" is reduced, but so is the consonant clarity and the air around it. On a bright vocal, this trade-off can sound worse than the original sibilance, just in a different way.

Split-band de-essers can help limit side effects, but they require careful setup for each source. In a busy session, you often don't have time for that level of detail, especially when working through 40 dialogue tracks from a location shoot with changing rooms and mic positions over several days.

Processing The Problem, Not The Region Around It

VoiceAssist's sibilance control works spectrally instead of using broadband gain reduction. It targets only the harsh parts in each detected spot, instead of lowering everything nearby in the frequency range. This keeps the presence and clarity on either side of the problem untouched. On a bright vocal or a clean dialogue track, this can mean the difference between a fix you can't hear and one that stands out to a careful listener.

When recording dialogue on location, mic positions and room acoustics often change between setups. This affects the sibilance, so using the same processing across the whole track will either miss some problems or overdo it in other spots.

VoiceAssist Shows You What You Are Working With

VoiceAssist highlights each detected sibilance region right on the waveform, and you can edit each one separately. You can adjust boundaries, change the intensity for certain sections, add missed detections, or remove false positives without affecting the rest. In an ARA workflow with Pro Tools, Cubase, or Fender Studio Pro, this automation stays with the clip and moves with the session, which is helpful when material passes between editors before mixing.

Of course, in some workflows, you want to print the changes, and this is also possible with VoiceAssist. In post-production, this workflow differs from using a de-esser as a real-time insert. Here, you make region-level decisions that you can see and change, instead of just setting a threshold and hoping it works for an entire reel of dialogue recorded in different conditions.

Each sibilance region is individually selectable. Adjust intensity, boundaries, or remove false detections.

Two Modes Per Region

VoiceAssist lets you use both volume reduction and spectral processing, and you can apply either one to specific regions instead of the whole track.

Volume reduction is quick for simple cases, while spectral processing works better for tough spots and keeps more of the original sound. Being able to choose per region lets you focus your efforts where they matter most, rather than treating everything the same way.

A De-Esser That's Not One Size Fits All

When working with comped vocals from different sessions or with small changes in mic position, the sibilance can change from phrase to phrase. A fixed threshold and band setting can't keep up without lots of manual tweaks. Spectral, region-level editing solves this because VoiceAssist analyzes each occurrence on its own instead of making one decision for everything.

For location dialogue, a character might be recorded indoors with reflections one day and outdoors with wind protection the next. The sibilance and mic proximity can change a lot across the scene, so you need a process that adapts to each case. Being able to review and adjust each detection means you don't have to rely on automation to work perfectly through a complex reel.

In bright vocal mixes where the vocal is high and open, sibilance often sits at the same frequency as important presence. Spectral processing lets you reduce harshness without pushing the vocal back in the mix, something a broadband de-esser usually can't do on this type of source.

Just Part Of The Power In VoiceAssist

VoiceAssist's sibilance control works together with denoising, dereverberation, dynamics riding, breath detection, and a content-aware gate, all in the same ARA session.

For sessions that need several of these tools, like most location dialogue and live vocal work, handling everything in one pass instead of using separate plugins can save a lot of prep time.