Tutorials Audio

Removing Wind Noise from Field Recordings in iZotope RX

Advanced · ~25 min

Overview

Wind noise on a field recording isn't one problem. It's usually two stacked on top of each other: a low-frequency rumble from the mic capsule moving in the airflow, and broadband turbulence noise that shares real spectral space with your subject. A single "reduce noise" slider treats both the same way and usually leaves you choosing between residual wind and a smeared, underwater-sounding result. This is the actual multi-stage workflow professional re-recording mixers use instead.

What You Need

iZotope RX (Standard covers most of this, De-wind specifically is an RX Advanced module)
The affected file, ideally with some clean "noise-only" handle at the head or tail for profiling
Reference monitoring you trust. This workflow is entirely ear- and eye-driven

Steps

Diagnose the noise in the spectrogram before touching a plugin

Switch to the Spectrogram view and look at the shape of the problem. Wind rumble shows up as a dense, low-frequency band that pulses with gusts. Broadband turbulence looks like fuzzy vertical smearing across a wider range. Knowing which one (or both) you're dealing with determines which module actually helps.

Clear sub-40Hz rumble with a high-pass filter first

Almost nothing you care about lives below 40-60Hz in dialogue or most sources. A steep high-pass here removes a large chunk of the rumble component for free, before any adaptive processing has to work on it, cheaper and safer than asking De-noise to do it.

Build a wind-specific noise profile for spectral de-noise

Select a section that's noise-only if you have one, or use De-wind's automatic profiling on Advanced. Learn the profile from a representative gust, not a quiet moment. The algorithm needs to see the actual character of the noise it's meant to remove.

Hand-repair remaining gusts with Spectral Repair

For isolated loud gusts that survive the profile-based pass, zoom into the spectrogram and paint directly over the offending energy with Spectral Repair in "Attenuate" or "Replace" mode. This is slower but far more surgical than pushing a global reduction amount higher.

A/B against the untreated file and back off if it sounds underwater

Toggle bypass constantly. The tell-tale sign of over-processing is a swirly, underwater or metallic quality on transients, if you hear it, reduce the amount or reduce-and-repeat in smaller passes instead of one aggressive pass.

Re-check the result in context, not solo

Processing that sounds acceptable soloed can reveal artifacts once it's sitting under music or next to cleaner takes. Always do the final pass/reject decision in the mix context, not in isolation.

Pro Tips

Process a duplicate track and keep the original untouched, spectral repair decisions are easy to regret a week later with fresh ears.
If the whole shoot was windy, profile once from the worst section and apply consistently rather than re-tuning per clip, inconsistent processing between cuts is more noticeable than a slightly imperfect but consistent one.
A dead cat or blimp on the mic during the shoot prevents 90% of this problem at the source. This workflow is for recordings you already have, not a substitute for field technique.

Knowledge Base

Why One Slider Doesn't Fix Wind Noise

Wind noise isn't a single, consistent signal the way tape hiss or a hum is: its energy content and level change gust to gust, and it overlaps directly with the frequency range of speech and most instruments. A single broadband reduction amount is always a compromise between under-treating the loudest gusts and over-treating the quiet moments in between.

Rumble vs. Turbulence

Low-frequency capsule rumble (wind physically moving the diaphragm) and broadband turbulence noise (air moving across the capsule) behave differently and respond to different tools. A high-pass filter handles the former almost for free, while the latter genuinely needs adaptive spectral processing.

Prevention Still Beats Repair

Every step in this guide is recovery, not prevention. A proper windscreen (foam for light breeze, a blimp/zeppelin with a fur windjammer for anything more) avoids most of this problem entirely, and no plugin fully recovers a gust that completely masked the source for an extended moment.

Where This Fits

This guide covers one specific part of audio recording. The wider picture, how microphone distance, room acoustics, polar patterns, and gain staging decide whether a recording is usable before any processing, is in Audio Recording Fundamentals: The Complete Guide, which frames the discipline as a whole and links out to the detailed guides underneath it, including this one. If you are starting from scratch rather than solving a specific problem, read that first and come back here.

FAQ

Q: Can RX remove wind noise from dialogue without hurting the voice?
A: Partially, and it's much harder than non-dialogue noise because wind energy and vowel energy overlap in the same low-mid frequency range. De-wind is tuned for this case specifically, but expect to trade some processing artifact for noise reduction on badly-affected dialogue. A dead cat or blimp on the day is still the real fix.

Q: Is a plugin ever going to fully rescue a take ruined by wind?
A: Not always. Sustained heavy gusts that fully mask the signal for more than a fraction of a second usually can't be recovered cleanly, spectral repair works by inferring missing content from the surrounding spectrum, and it runs out of information to infer from past a certain point.