ICEPICK
Manual de-essing. Reimagined.
Event-based de-essing for vocals. Each sibilant, plosive, and breath is detected individually. Nothing is applied until you decide.
Intelligent starting points. Your finishing touches.
Precision. Without fatigue.
Fly through detections. At modern speeds.
Trained on the human voice. Built for engineers.
IcePick.
Available as an early access release.
Introductory pricing during beta.
Try IcePick14-day trial · macOS · AU / VST3
Why we built IcePick
We've tried every de-esser, and the result is always the same. At some point, we revert to manual de-essing — sometimes to finish the job, often to correct what automation couldn't. Rule-based de-essers can't resolve the tradeoff: set the threshold low enough to catch everything and the harshest sibilants turn lispy; set it higher and quieter but still problematic events slip through.
Manual de-essing has always been the solution. It offers full control, event by event. Every sibilant (s, sh, ch), hard consonant (t, k, p, b, d), and breath has to be found, shaped, and balanced by hand — one decision at a time. The results are worth it. The process isn't sustainable.
IcePick exists to make that level of control practical. It removes the time cost without removing authority. IcePick is currently available as an early access release, refined through real-world use. Early users help shape what it becomes, and pricing will increase after launch.
If you want your voice to inform its development, you're invited to contribute material to the detection model. Submitted vocals are reviewed and edited by hand, and used to improve how IcePick responds to real voices.
Early access updates
Updates on development, early features, and early access pricing.
FEATURES
Detection Modes — Sibilance, hard consonants, breath
Per-Event Editing — Reduction, boundaries, mute state
Batch Operations — Multi-select, apply edits simultaneously
Crossfade Control — Configurable fade-in and fade-out
SPECIFICATIONS