Izotope Rx Guide

Covers iZotope RX 11 — Updated May 2026

No other piece of software has saved more recordings, rescued more vocal takes, and salvaged more nearly-finished albums than iZotope RX. Originally built for broadcast and film post-production, RX has become an essential part of the modern music production toolkit — and for good reason. Whether you are dealing with a buzzing guitar amp, a vocalist who recorded in a room with a noisy HVAC system, a vinyl sample crammed with crackle, or a live recording marred by crowd noise, RX has a dedicated, purpose-built tool to handle it.

This guide covers every major module in iZotope RX from a music producer's perspective, with practical settings, real-world workflow examples, and honest guidance on what RX can and cannot do. We will also cover how RX integrates into your existing DAW workflow via RX Connect and ARA2, so you can use it without disrupting your session flow.

iZotope RX Versions and Editions: Which One Do You Need?

iZotope offers RX in three tiers: RX Elements, RX Standard, and RX Advanced. There is also a standalone version of the RX application that ships with all paid tiers. Understanding which version you need is the first decision you will make.

For most music producers, RX Standard is the sweet spot. It gives you Spectral Repair, De-reverb, full De-noise, and Ambience Match — the modules you will reach for most often on music sessions. RX Advanced adds Music Rebalance and Dialogue Isolation, which are genuinely useful if you work with stems, remixes, or mixed recordings where you need to isolate or adjust individual sources after the fact. If you are a casual user who just wants to clean up home recordings, RX Elements handles the basics at an accessible price point.

iZotope frequently bundles RX with other tools in their Music Production Suite or Post Production Suite, so check bundle pricing before buying individually. Sales are common, and prices fluctuate significantly around major promotional periods.

RX Workflow: Standalone App vs. RX Connect vs. ARA2

One of the first things that confuses new RX users is how to actually use it in their productions. Unlike most plugins, iZotope RX is primarily a standalone application — a fully featured audio editor with its own waveform display, spectral display, transport, and processing engine. You open audio files directly in RX, process them, and then re-import the results into your DAW. But this is only one of three workflows available to you.

The Standalone Application

The RX standalone app is a purpose-built audio editor with a dual-mode display: a standard waveform view on top and a spectrogram view below (or in full-screen spectral mode). The spectrogram is color-coded by frequency and amplitude, with time running left to right and frequency running bottom to top — low frequencies at the bottom, high frequencies at the top. Bright regions indicate louder energy at that frequency. This visual representation is what makes RX so powerful for identifying and repairing specific sonic problems.

To use the standalone app in a music production context, you typically:

1. Export the problematic audio region from your DAW as a wav or aiff file
2. Open it in RX
3. Identify and repair the issue using one or more modules
4. Export the processed file back to disk
5. Re-import it into your DAW session

This round-trip workflow is precise but slow, especially if you are iterating on repairs. For most music production scenarios, RX Connect is a better option.

RX Connect: The Round-Trip Plugin

RX Connect is a plugin that lives inside your DAW — it appears in your plugin menu as a standard insert effect. When you click "Send to RX" in the RX Connect plugin, it automatically transfers the audio from the selected clip into the RX standalone application. You make your repairs there, then click "Send Back" and the processed audio drops back into your DAW session, perfectly aligned to the original position in the timeline. This eliminates the manual export/import step and makes the RX workflow feel much more integrated.

RX Connect works in Pro Tools (via AudioSuite), Logic Pro, Ableton Live, FL Studio, Studio One, and most other major DAWs. In Pro Tools, you can use it as an AudioSuite plugin for offline processing. In DAWs that support ARA2 — including Studio One, Logic Pro, and Reaper — you have the most seamless experience of all.

ARA2 Integration

ARA2 (Audio Random Access version 2) is a plugin communication standard that allows certain plugins to access the full audio context of a clip rather than just the real-time audio stream. With ARA2, iZotope RX integrates directly into the DAW editing environment — you can open RX processing on a clip without leaving your DAW, and changes reflect immediately without a separate send/receive step. This is the fastest and most modern workflow for RX users whose DAW supports ARA2.

Noise Reduction: De-noise, Voice De-noise, and Dialogue Isolation

Noise reduction is the most commonly used function in iZotope RX for music producers, and it comes in several forms. Understanding which tool to reach for — and how hard to push it — is what separates professional-sounding repairs from obvious, artifact-riddled processing.

Spectral De-noise

The main De-noise module in RX uses a technique called spectral subtraction combined with machine learning-based noise profiling. The basic approach is: you select a region of audio that contains only the noise you want to remove (a "noise print"), tell RX to learn what that noise sounds like, and then it subtracts a model of that noise from the full recording.

The key controls in De-noise are:

• Reduction: How much of the profiled noise to subtract, measured in dB. For music recordings with light room noise, 6–10 dB of reduction is usually enough. Pushing to 20+ dB will start to cause musical artifacts — a swirling, underwater quality called "musical noise" that is often more distracting than the original noise floor.
• Threshold: Determines at what level the noise reduction kicks in. Higher thresholds preserve more of the signal at the cost of less aggressive noise removal.
• Artifact Control: A smoothing parameter that trades artifact reduction for processing speed. For music (especially pitched material), a higher Artifact Control value helps avoid the metallic warbling that can occur with aggressive settings.
• Learn (Adaptive Mode): In adaptive mode, RX continuously analyzes the noise floor and updates its model in real time. This is useful for recordings where the noise character changes over time — like a room with air conditioning that cycles on and off.

Practical tip: For music recordings, always prefer the minimum amount of De-noise that makes the problem tolerable rather than trying to get a perfectly clean result. A recording with 8 dB of noise reduction applied conservatively sounds far more natural than one where 20 dB has been applied aggressively. If you need to share the mix with collaborators and want guidance on how to mix vocals professionally, noise reduction is just one piece of the puzzle — eq, compression, and room treatment all matter more in the long run.

Voice De-noise

Voice De-noise is a machine learning-based module introduced in RX 7 that takes a fundamentally different approach. Rather than building a noise model from a selected region, it uses a neural network trained on thousands of hours of speech and vocal audio to separate voice from background noise in real time, without requiring a noise print at all.

Voice De-noise has a single primary control: a Noise Reduction slider from 0–100%. For music vocals, the sweet spot is usually between 30–60%. Above 70–80%, you start to hear the tell-tale artifacts of neural noise reduction — a kind of bubbling or metallic quality around consonants and breath sounds. The module also has a Voice Sensitivity control that determines how aggressively it classifies audio as "voice" versus "noise". On melodically sung vocals with extensive vibrato or falsetto, reducing Voice Sensitivity slightly can help preserve more of the vocal character.

Voice De-noise is the fastest tool in RX for cleaning up home studio vocal recordings, podcast recordings, and any situation where someone sang or spoke into a microphone in a less-than-ideal acoustic environment. It is not a magic bullet, but for light-to-moderate room noise on vocal tracks, it is remarkably effective with very few controls to learn.

Dialogue Isolation (RX Advanced Only)

Dialogue Isolation is the most powerful voice-isolation tool in the RX suite. It uses a more sophisticated neural network architecture than Voice De-noise and is capable of separating voice from complex, non-stationary noise — crowd noise, music, traffic, HVAC systems with varying pitch, and more. For music producers, Dialogue Isolation is most useful when you have a vocal recording that was captured in a problematic environment and De-noise or Voice De-noise cannot fully resolve it. It is also increasingly used for sample processing — extracting a vocal from an old recording with poor technical quality.

The trade-off with Dialogue Isolation is that it requires RX Advanced and can introduce its own artifacts at high reduction amounts. Use it as a last resort after De-noise and Voice De-noise have been pushed to their limits. The module processes audio in blocks and does not require a noise print — just set the reduction amount and listen critically.

Spectral Repair, De-click, and De-crackle

Beyond noise reduction, some of the most musically useful modules in RX are the tools designed to repair transient-level events: clicks, crackles, pops, digital clipping artifacts, and more. These are the kinds of problems that show up as spikes in the waveform view or as bright vertical lines in the spectrogram.

Spectral Repair

Spectral Repair is one of the most impressive and uniquely RX capabilities in the entire suite. It works by selecting a problem region in the spectrogram view — using the lasso, rectangle, or brush tool — and then using the surrounding audio context to synthesize replacement audio that fills in the selected region. Think of it as the audio equivalent of Photoshop's content-aware fill, but for sound.

Spectral Repair has four processing modes:

• Attenuate: Reduces the level of the selected region without replacing it. Useful for persistent tonal artifacts like a hum partial or a resonant squeak that you want to reduce rather than eliminate.
• Replace: Synthesizes new audio to fill the selected region based on the audio immediately before and after it. Best for short, discreet artifacts — a click, a digital glitch, a cough — where the surrounding audio is clean and provides enough context for a convincing reconstruction.
• Interpolate: Like Replace, but uses audio from both before and after the selection and blends them together. This tends to produce the most natural results for tonal material like sustained notes and vowel sounds.
• Pattern: Analyzes a pattern from a separately selected reference region and uses that pattern to reconstruct the selected region. Useful for highly repetitive audio content like drum loops with periodic artifacts.

In music production, Spectral Repair excels at tasks like: removing a single loud cough from a live recording without muting the surrounding performance, erasing the click of a microphone stand knock, repairing digital corruption in a sample, removing a brief squeak from a guitar string, and cleaning up clicks and thumps in vinyl samples. For producers working extensively with samples, Spectral Repair combined with De-click is the fastest path to clean, usable source material.

De-click and De-crackle

The De-click module is designed for automatic removal of impulsive noise — clicks, pops, and other broadband transient artifacts that are time-limited and not musically intended. It works by analyzing the statistics of the audio signal and identifying events that are statistically anomalous — too loud, too brief, too broadband to be part of the intended signal — and interpolating them out.

Key controls for De-click:

• Sensitivity: Higher values detect more artifacts but also increase the risk of false positives — removing legitimate transients like drum hits, guitar picks, or piano attacks. For music, keep Sensitivity at 40 or below unless you are working on material without sharp transients.
• Frequency Skew: Adjusts the frequency range where the algorithm looks for clicks. A Treble setting focuses on high-frequency clicks typical of vinyl. A Multi setting addresses clicks across all frequencies. For digital artifacts, use Multi.
• Click Widening: Determines how wide a region around each detected click is interpolated. Keep this narrow for music with sharp transients.

De-crackle targets the lower-level, higher-density crackle typical of vinyl records, old tape recordings, and deteriorated digital media. It is essentially a high-density version of De-click, optimized for the statistical signature of continuous low-level crackle rather than isolated clicks. For lo-fi production and sample flipping, De-crackle is invaluable — it lets you add the character of vinyl (via a stylus, record player, or physical pressing) without the distracting high-frequency crackle that makes samples hard to layer. If you are interested in that aesthetic, check out our guide on how to make lo-fi hip-hop for context on when to preserve versus remove vinyl artifacts.

De-clip

Clipping happens when a signal exceeds the maximum amplitude that the recording medium or converter can represent — the waveform is literally cut flat at the top and bottom, replacing the natural peaks with a hard, horizontal line. The sonic result is distortion — a harsh, buzzy quality that is immediately obvious to most listeners and nearly impossible to fix with standard mixing tools.

RX's De-clip module uses waveform reconstruction to analyze the shape of the clipped peaks and estimate what the original waveform should look like — then synthesizes that missing information. It is not perfect, especially for heavily clipped material (more than 6 dB of clipping), but for moderate clipping — the kind that happens during a passionate vocal performance or a drummer who hits harder than expected — it can produce dramatic improvements.

The main controls are Clip Threshold (the level at which clipping is detected — leave this on Auto unless you have a specific reason to set it manually) and Algorithm, which offers Standard and Declip Clipper options. For music, Standard is appropriate in most cases. Always apply De-clip before any other processing — running noise reduction or spectral repair on a clipped signal will embed the clipping artifacts into the processed result.

Music Rebalance, Ambience Match, and De-reverb

The modules in this section represent RX's most ambitious audio intelligence capabilities — tools that attempt to understand the musical content of a recording and manipulate it in ways that would have been impossible even a decade ago.

Music Rebalance (RX Advanced)

Music Rebalance is one of the most remarkable tools in the RX Advanced package. It uses source separation technology — similar to, but more conservative and musically aware than, dedicated AI stem separation tools — to identify and independently adjust the levels of four stems within a mixed recording: Vocals, Bass, Percussion, and Other (which covers everything else: guitars, keys, strings, and so on).

The practical applications for music producers are significant:

• Reducing the level of a bass guitar in a mixed stem when the master fader is already at unity and the bass is overwhelming
• Boosting the percussion stem in a mixed drum submix for a more aggressive feel without re-mixing from scratch
• Attenuating the vocal in a karaoke-style instrumental extract
• Rebalancing a sample so that the original bass does not fight with your new arrangement
• Extracting usable stems from a fully mixed track for remixing purposes

Music Rebalance is not a perfect stem separator — it works best on well-produced, commercially mixed material with clear separation between sources. On dense, highly compressed material, the stems bleed into each other and the results can sound smeared or artifact-laden. It is also worth noting that the Other stem is the hardest to isolate cleanly, because it is a catch-all for everything that is not vocals, bass, or percussion.

For producers who want purpose-built, AI-powered stem separation rather than the more conservative Music Rebalance approach, there are dedicated tools available — our guide to AI stem separation tools covers the current landscape in detail.

Music Rebalance's controls are straightforward: four sliders (Vocals, Bass, Percussion, Other) ranging from -6 dB to +6 dB. There is also a Solo button for each stem so you can audition the separation quality before committing. As a rule, adjustments of ±3 dB or less produce the most transparent results; larger adjustments increase the audibility of separation artifacts.

Ambience Match

Ambience Match solves a specific and extremely common music production problem: you have two recordings that were captured in different acoustic environments, and they need to sit together in the same track or album without sounding discontinuous. Maybe a vocal overdub was recorded in a different room from the original session. Maybe you are editing together two takes from different microphone positions. Maybe you are adding a new voiceover to a piece of music that already has an established room sound.

Ambience Match analyzes the ambient noise and early reflections of two recordings — a "target" (the sound you want) and a "source" (the sound you want to transform) — and applies the ambience character of the target to the source. It is not a convolution reverb. It analyzes the real noise floor and room character of both recordings and creates a statistical match, which means it can also add a specific noise floor texture to a clean recording to make it blend seamlessly into a noisier environment.

For music producers, Ambience Match is most useful in editing heavy sessions where continuity between takes matters — live recordings, vocal compilations, and album projects where multiple recording environments are stitched together. It is also useful for sample work, helping a synthesized or processed element blend into a recording that has a distinct acoustic character.

De-reverb

De-reverb is one of the most eagerly awaited and frequently requested audio repair tools ever developed, and RX's implementation is genuinely impressive given the difficulty of the problem. It uses a combination of spectral analysis and machine learning to identify and reduce the reverberant energy in a recording — essentially, it tries to "dry out" a reverb-heavy recording and recover something closer to the dry source.

The key controls are:

• Reduction: The amount of reverb reduction, from 0–100%. For music, 20–40% is usually the practical ceiling before artifacts become distracting. Beyond that, the dry sound can start to take on a hollow, phasey quality.
• Tail Length: How long the reverb tail is estimated to be. Match this to the actual room character of the recording for best results.
• Artifact Control: Reduces smearing and transient artifacts at the cost of slightly less aggressive reverb reduction.

It is important to set realistic expectations about De-reverb. It cannot completely remove a heavily reverberant recording — if the original signal was captured in a large, live room with 3 seconds of reverb, De-reverb can reduce the reverb but cannot produce a recording that sounds like it was captured in a vocal booth. For moderate reverb situations — a bedroom recording with too much room sound, a vocal captured too close to a reflective wall — it can make a meaningful improvement. After applying De-reverb, a modest amount of added reverb from a high-quality algorithmic reverb can help restore natural-sounding space without the original room's problematic character. For more on this, see our guide on how to use reverb on vocals.

Spectral Editing, EQ Module, and Advanced Repair Techniques

Beyond the dedicated repair modules, the RX standalone application contains a full spectral editor that allows you to make surgical selections in the frequency-time domain and apply processing or manual corrections that are simply impossible in standard DAW waveform editors.

Spectral Editing Fundamentals

The spectral display in RX shows audio as a two-dimensional grid where the horizontal axis represents time (left to right) and the vertical axis represents frequency (bottom to top). The color of each point in the grid represents the energy at that frequency at that moment in time — typically from dark blue (quiet) through green and yellow to red and white (loud).

You can select regions in this spectrogram using four selection tools:

• Time Selection: Selects across all frequencies for a given time range. Equivalent to a standard waveform selection.
• Frequency Selection: Selects all time within a specific frequency range — useful for targeting a specific hum frequency or tonal artifact that persists across the whole clip.
• Rectangle Selection: Defines a specific time-frequency rectangle. The most common selection type for targeted repairs.
• Lasso Selection: A freehand selection tool. Allows highly specific, irregular selections around complex artifacts in the spectrogram.

Once you have made a selection, you can apply any of RX's modules, fade the selection, copy it, or manually attenuate it. The Lasso selection combined with Spectral Repair Replace is one of the most powerful repair workflows in RX — you can literally draw around a problematic sound artifact in the spectrogram and surgically reconstruct that specific frequency-time region without affecting the surrounding audio at all.

Hum Removal

The De-hum module is dedicated to removing electrical hum — the 50 Hz or 60 Hz (and associated harmonics) that comes from ground loops, inadequate shielding, poorly designed cables, or proximity to electrical equipment. For producers who record live instruments, especially electric guitars and basses, hum is a persistent reality.

De-hum's key control is Hum Frequency — set this to 50 Hz if you are in Europe/Asia (where mains electricity is 50 Hz) or 60 Hz if you are in North America. RX can auto-detect the fundamental hum frequency if you are unsure. The Number of Harmonics control determines how many harmonic overtones of the fundamental frequency are attenuated. Electrical hum is rich in odd harmonics, so using 5–10 harmonics typically provides a complete removal. The Reduction slider controls how much attenuation is applied at each harmonic frequency.

De-hum is often more effective than a notch filter in an EQ because it tracks the exact frequency and harmonic structure of the hum without requiring you to manually set multiple notch filters. For producers dealing with consistent hum on guitar recordings, De-hum can make the difference between a usable and unusable take.

The EQ Module

RX includes its own EQ module, which at first glance seems redundant given the excellent standalone EQ plugins available. But RX's EQ has a unique advantage: it operates in the spectral domain on offline audio, meaning it can produce zero-latency, phase-linear results with extremely high frequency resolution. It is also tightly integrated with the spectral display — you can set EQ points directly by clicking on features in the spectrogram that you want to target.

For most mixing tasks, you will use a dedicated mix-bus or channel EQ plugin — our roundup of the best EQ plugins for mixing covers all the top options. But for repair work — correcting a specific resonance in a recording, notching out a persistent tonal artifact, or preparing audio for stem separation — RX's EQ is a genuinely useful tool.

Loudness Control

The Loudness Control module in RX Standard and Advanced allows you to measure and normalize the integrated loudness of audio files to LUFS targets used by streaming platforms — specifically -14 LUFS (Spotify, Apple Music, YouTube), -16 LUFS (podcasts), and custom user-defined targets. For music producers delivering stems, albums, or singles for streaming, Loudness Control provides a quick way to check and adjust loudness compliance without opening a separate metering application.

That said, Loudness Control is not a substitute for proper mastering — it is a measurement and gain adjustment tool, not a dynamic processor. If you are applying loudness normalization to unmastered material, be aware that the process does not compress or limit the audio — it simply adjusts the overall gain to hit the target LUFS level. For proper mastering workflows, see our iZotope Ozone 12 review which covers the full mastering suite that pairs naturally with RX.

RX in Real Music Production Workflows

Understanding the individual modules is important, but understanding how to apply them in realistic production scenarios is what makes the difference between knowing about RX and actually using it effectively. Here are the most common music production scenarios where RX is the right tool, along with recommended workflow approaches.

Vocal Recording Repair

This is the number one use case for music producers. A vocalist records a take that is musically excellent but technically compromised — room noise, HVAC hum, clothing rustle, mouth noise, a chair creak at a critical moment. Before committing to using a problem take, send it to RX for triage:

1. Open the file in RX via RX Connect or the standalone app
2. Zoom into the spectrogram and identify the problem types: broadband noise (use De-noise or Voice De-noise), tonal hum (De-hum), isolated transient artifacts (De-click or Spectral Repair), clipping (De-clip first)
3. Apply De-clip if any clipping is present — always first
4. Apply Voice De-noise at a conservative setting (30–50%) to reduce broadband noise
5. Use De-hum if there is obvious 50/60 Hz hum
6. Use Spectral Repair with Rectangle or Lasso selection to eliminate specific isolated problems (a cough, a chair creak, a single click) that the automated modules did not catch
7. A/B the repaired file against the original. If the repair sounds worse than the problem, reduce the processing
8. Send back to DAW and continue mixing

For a comprehensive look at the vocal mixing chain that follows repair, see our guide on advanced vocal mixing techniques.

Sample Preparation and Flipping

Producers who work with samples — vinyl records, old cassettes, obscure digital files, archival recordings — regularly encounter audio that has technical problems baked into the source material. RX is the premier tool for sample preparation:

• Vinyl samples: Use De-click (Sensitivity 25–40, Treble or Multi frequency skew) to remove surface noise, then De-crackle (Low to Medium setting) for the crackle texture. If the record has hum from a poorly grounded turntable, apply De-hum first.
• Old tape samples: Tape often has broadband hiss. Take a noise print from a silent section between audio events and apply De-noise. Tape also degrades in ways that can cause subtle wow and flutter — RX's Pitch Contour module can help with this in advanced cases.
• Lo-fi aesthetics: Paradoxically, many lo-fi producers use RX to selectively clean up samples just enough to make them layerable, while preserving enough character to retain the vintage feel. The goal is not a perfectly clean recording — it is controlled, usable imperfection.

Live Recording Editing

Live recordings present unique audio repair challenges because the noise environment is complex and varies over time — crowd noise changes between songs, environmental acoustics shift as the room fills up, and individual audience events (coughs, phones, conversations) can punctuate otherwise excellent performances. RX handles this environment better than any other tool:

• Use adaptive De-noise mode to track changing noise floor characteristics throughout a long performance
• Use Spectral Repair with Lasso selection to remove individual crowd events (a loud cough during a quiet passage, a ringing phone during a piano solo)
• Use Ambience Match to smooth over continuity differences between separately recorded sections or nights of a multi-night run
• Use De-reverb conservatively on particularly boomy or echoey recordings to improve clarity before mixing

Podcast and Voiceover Integration

While this is primarily a music production guide, many music producers also record or produce podcasts, voiceovers, and content that sits alongside music. RX's Voice De-noise and Dialogue Isolation modules are the industry standard for this use case — they appear in virtually every professional podcast post-production chain. If you produce both music and spoken word content, RX Standard or Advanced covers both domains comprehensively. For producers interested in the podcast recording side, our guide on how to record a podcast covers the full setup workflow.

Batch Processing with RX Monitor

For high-volume work — preparing large sample libraries, cleaning up entire albums of archival recordings, or processing multiple podcast episodes — RX's batch processing capabilities are essential. You can build a module chain (De-clip → De-noise → De-click → Loudness Control, for example) and apply it to an entire folder of audio files automatically. This is available through RX's standalone application File menu and is a significant time-saver for producers who regularly handle large amounts of audio material.

The batch processor supports processing chains of any length, custom LUFS targets for loudness normalization, multiple output formats (wav, aiff, mp3, flac), and bit depth/sample rate conversion. For sample pack creators, batch processing an entire library through a consistent De-click and normalization chain before delivery is considered professional standard practice.

Common Mistakes, Best Practices, and What RX Cannot Do

Even experienced users make mistakes with RX that lead to artifacts, diminished recording quality, or unnecessary time spent on processing that cannot actually solve the underlying problem. Here is what to know before you reach for any of the tools.

The Most Common Mistakes in RX

1. Applying noise reduction to an already-mixed track instead of individual stems
Noise reduction algorithms are trained and optimized to work on single-source or simple recordings. Applying aggressive De-noise to a fully mixed two-track stereo file will affect the desired audio content — particularly instruments with similar spectral characteristics to the noise being removed — and degrade the mix quality. Whenever possible, apply RX processing at the stem or individual track level before mixing.

2. Using RX as a substitute for good recording practice
RX is a repair tool, not a preventive one. A recording made in a properly treated room with good microphone technique and appropriate gain staging will always sound better than a poorly recorded take processed through every module in RX. Learning to control your recording environment, choosing the right microphone for the source, and setting appropriate gain will save you far more time than learning to fix problems after the fact.

3. Over-processing
This is by far the most common mistake. Every module in RX, when pushed too hard, introduces its own category of artifact. De-noise produces musical noise. Voice De-noise produces bubbling and metallic consonants. De-reverb produces phasiness and hollowness. De-click can eat legitimate transients. The rule of thumb: process to the minimum level that makes the problem acceptable, not to the maximum level that eliminates it. A slightly noisy recording that sounds natural is almost always preferable to an artifact-free recording that sounds processed.

4. Not listening critically on multiple systems
Artifacts from RX processing — especially from De-noise and De-reverb — can be difficult to hear on studio monitors in an acoustically treated room but obvious on earbuds, laptop speakers, or consumer headphones. Always check processed audio on multiple playback systems before committing. See our guide on how to make music that translates on any system for a complete monitoring workflow.

5. Forgetting to apply De-clip before anything else
De-clip needs access to the clipped waveform in its undamaged context. If you apply De-noise first and it modifies the signal, De-clip will have a harder time reconstructing the original peaks accurately. Always De-clip first.

What RX Genuinely Cannot Fix

Setting realistic expectations about RX's capabilities is important for avoiding frustration and wasted time. There are things RX simply cannot do:

• Severely clipped recordings (6+ dB of clipping): De-clip can repair moderate clipping well, but when the peaks have been so severely truncated that the original waveform shape cannot be estimated from context, the reconstruction will sound audibly wrong.
• Completely buried signals: If the signal-to-noise ratio of the original recording is below about -10 dB (the wanted sound is more than 10 dB quieter than the noise), no noise reduction algorithm will recover a clean signal. The wanted audio simply does not contain enough information to separate from the noise.
• Codec or compression artifacts from low-bitrate encoding: MP3 or AAC artifacts — the warbling, swirling metallic quality of heavily compressed audio files — are fundamentally different from recording noise or environmental contamination. While RX has some tools that can slightly reduce these artifacts, it cannot restore the frequency and dynamic information that lossy compression permanently discards.
• Phase cancellation problems: If two microphones captured the same source and the resulting recording has significant phase cancellation (comb filtering), RX does not directly address this. Phase correction requires dedicated phase alignment tools or manual EQ correction.
• Room acoustics problems in live mixing: RX is an offline or near-real-time processing tool for recorded audio. It is not a live room correction system. If you have acoustic problems in your monitoring environment, the solution is acoustic treatment — our guide on home studio acoustic treatment covers this comprehensively.

Best Practices Summary

To use RX most effectively in music production:

• Work at the individual track or stem level, not on the master mix
• Process in the correct order: De-clip → De-hum → De-noise/Voice De-noise → Spectral Repair for isolated artifacts → De-reverb (if needed)
• Use conservative settings and A/B constantly — always compare processed against unprocessed
• Keep the standalone app open during mix sessions for quick access via RX Connect
• Save your noise prints and module presets for recurring recording environments
• Batch process when handling large volumes of similar audio material
• Know when to re-record — if RX cannot fix it cleanly, the right answer is often another take

iZotope RX is updated regularly, with iZotope releasing significant new features and AI model improvements approximately annually. This guide covers RX 11 features and behavior as of May 2026. Always check the iZotope release notes for changes to specific module behaviors in newer versions, as the AI-powered modules in particular can change substantially between major releases.

Practical Exercises

Frequently Asked Questions