iZotope RX Spectral Repair: Step-by-Step Cleanup Guide

You’ve spent hours in a dark location, recorder running, asking questions into the silence. Back home, you play the audio and hear something, a faint whisper buried under hum, static, or wind noise. It could be significant evidence, or it could be nothing. The only way to find out is to clean up that recording without destroying what’s underneath. That’s exactly where iZotope RX Spectral Repair comes in. It’s one of the most precise audio restoration tools available, and for paranormal investigators working with noisy EVP captures, it can mean the difference between a compelling piece of evidence and an unlistenable mess.

At Haunt Gears, we spend a lot of time talking about the hardware side of investigations, EMF meters, thermal cameras, recorders. But gear is only half the equation. Post-investigation audio processing is where raw recordings turn into analyzable evidence, and having the right software matters just as much as having the right microphone. iZotope RX has become a go-to tool among serious investigators who want to isolate unexplained sounds while maintaining the integrity of their original recordings.

This guide walks you through how to use the Spectral Repair module step by step. We’ll cover what it actually does, when to use it over other RX modules, and how to apply it to your investigation audio without introducing artifacts or accidentally wiping out the very sounds you’re trying to hear. Whether you’re new to audio editing or you’ve been running cleanup on recordings for years, this breakdown will give you a practical, repeatable workflow you can use after every investigation.

What Spectral Repair is and what it fixes

iZotope RX Spectral Repair is an audio restoration module that operates on a spectrogram view of your recording, a visual representation of frequency over time. Instead of applying broad filters across an entire track, it lets you target specific moments and frequency ranges for removal or reconstruction. Think of it as a content-aware healing tool for audio: you select a damaged region, and the module analyzes the surrounding sound to intelligently fill that gap or reduce the intrusion.

Spectral Repair works by using the audio surrounding your selection as reference data, which means it reconstructs sound based on what your recording actually contains rather than applying a generic noise curve.

How the reconstruction process works

When you make a selection in the spectrogram, Spectral Repair analyzes audio on both sides of that selection to build a model of what "should" be there. It then applies one of four modes to either replace, attenuate, or blend the targeted region. The software treats audio as a two-dimensional dataset, where horizontal position represents time and vertical position represents frequency, allowing surgical precision that waveform-only editors cannot match.

What kinds of sounds it targets

This module handles a specific category of problems: isolated, time-limited noises that appear within a longer recording. Common targets include microphone handling bumps, short interference bursts, clicking sounds from equipment, sudden wind gusts, and electronic glitches. For paranormal audio work, it’s particularly useful for removing a door creak or a recorder click that occurs right over a potential EVP response.

• Microphone handling noise
• Short electrical interference spikes
• Chair scrapes or body movement sounds
• Camera shutter clicks
• Brief wind bursts or breath artifacts

It will not fix continuous broadband noise like room hum or HVAC rumble. For those problems, RX’s Noise Reduction module is the better starting point.

Step 1. Get audio into RX and find the noise

Before iZotope RX Spectral Repair can do anything useful, you need to load your recording correctly and orient yourself within the spectrogram view. Open RX, then drag your audio file directly into the waveform display, or go to File > Open and navigate to your recording. RX supports WAV, AIFF, FLAC, and MP3, though working from an uncompressed WAV gives you the cleanest source material to analyze.

Navigating the Spectrogram Display

Once your file loads, switch the view from the standard waveform to the spectrogram display by clicking the spectrogram icon at the top of the editor. You will see frequency plotted vertically (low at the bottom, high at the top) and time running left to right. Brighter colors indicate louder energy at that frequency and moment.

Set your spectrogram resolution to at least 4096 FFT size in RX’s display settings to get enough detail for precise selection work on EVP recordings.

Use these keyboard shortcuts to move through the file efficiently:

• Spacebar: Play/pause
• Ctrl+Scroll: Zoom in on time axis
• Alt+Scroll: Zoom in on frequency axis
• R: Switch to the selection tool

Step 2. Select the unwanted sound in the spectrogram

With your file loaded and your problem area located, your next task is making an accurate selection around the unwanted sound. Precise selection is the most important skill in iZotope RX Spectral Repair work because the module only processes what you highlight. A sloppy selection either leaves noise behind or damages audio you want to keep.

Choosing the Right Selection Tool

RX gives you three selection tools, and picking the correct one changes your results significantly. Use these depending on the situation:

• Time selection (T): Selects across all frequencies within a time range. Best for short, full-spectrum bursts like a click or thump.
• Frequency selection (F): Selects across the full recording within a frequency band. Use this for a narrow hum at a fixed pitch.
• Lasso tool (L): Freehand selection for irregular noise shapes like a scrape or resonant interference spike.

Making a Precise Selection

Zoom in on the time and frequency axes before drawing your selection so you can see the exact edges of the noise event. Drag your selection tool to cover only the colored region representing the problem, leaving clean audio on both sides.

Always leave at least 0.1 seconds of clean audio on each side of your selection so the module has enough reference material to reconstruct the region accurately.

Step 3. Pick a mode and dial in settings

Once your selection is in place, open the Spectral Repair module from the left panel in iZotope RX. You will see four modes at the top: Attenuate, Replace, Partial, and Pattern. Each one processes your selection differently, and picking the wrong mode on a delicate EVP recording can introduce audible artifacts that weren’t there before.

The Four Modes Explained

Here’s how each mode behaves so you can choose the right one for your situation:

Mode	What it does	Best for
Attenuate	Reduces noise without full removal	Soft interference blended with wanted audio
Replace	Reconstructs the region from surrounding audio	Short clicks, thumps, or pops
Partial	Blends replacement with the original signal	Moderate noise over complex audio
Pattern	Matches patterns from elsewhere in the file	Repeating tonal interference

Adjusting the Settings

After selecting a mode, Bands and Surrounding Regions are the two main controls to set. Bands determines how many frequency divisions the module analyzes, and higher values give more precision at the cost of processing time. Start with Bands at 128 and Surrounding Regions at 3 for most investigation audio, then increase Bands to 256 if the repair sounds unnatural on playback.

For recordings with faint responses buried under noise, use Replace mode with Bands set to 256 to preserve the delicate frequency content beneath the problem region.

Step 4. Render, check results, and commit edits

With your mode selected and settings dialed in, click Preview to hear how iZotope RX Spectral Repair will process the selection before committing anything. Preview runs the repair in real time without writing any changes to your file, so you can adjust Bands or switch modes freely until the result sounds right.

Always preview before rendering because applying and undoing multiple renders on a delicate EVP recording can degrade audio quality over time.

Check the Repair on Playback

Once the preview sounds clean, click Render to apply the repair. Your selection will update immediately in the spectrogram display, and the treated region will show reduced or replaced energy where the noise used to appear.

Listen for these problems before you commit:

• Unnatural silence or hollow tone where the noise was removed
• Smearing or reverb-like artifacts on transients near the selection edges
• Sudden volume shift at the start or end of the selection boundary

If you hear any of those issues, press Ctrl+Z to undo, adjust your Bands value or tighten your selection boundary, and preview again. Once satisfied, go to File > Save As and save the repaired version under a new filename so your original recording stays intact as a backup.

Conclusion

iZotope RX Spectral Repair gives you a repeatable, precise process for cleaning up investigation recordings without guessing or applying broad filters that could erase the exact sounds you need to hear. You now have a complete workflow: load your audio, locate the problem in the spectrogram, select it accurately with the right tool, choose the mode that fits the situation, preview before rendering, and save a separate clean version. That sequence works on everything from a recorder click that lands on a potential EVP response to wind interference that blankets a full section of audio.

Your recording equipment and your post-processing workflow belong in the same conversation when you’re building out a serious investigation setup. Good gear captures clean source material, and software like RX helps you extract what’s actually on it. If you’re looking to upgrade the hardware side of your kit, browse the paranormal investigation equipment at Haunt Gears to find tools built for serious fieldwork.