Best Noise Reduction Plugins

Updated May 2026 — Noise is the enemy of every recording, from the low hum of a poorly grounded guitar amp to the HVAC rumble bleeding into a vocal take, from the tape hiss on a vintage sample to the broadband noise floor of an affordable condenser mic in an untreated room. No matter how good your signal chain is, unwanted noise finds a way into your recordings. Noise reduction plugins are the surgical tools that let you remove that noise without destroying the musical content you worked so hard to capture.

The plugin market for noise reduction has expanded dramatically over the past few years, driven by advances in machine learning and spectral processing. What used to require an expensive hardware unit or a dedicated workstation running iZotope RX in offline mode can now happen in real time, inside your DAW session, with adaptive algorithms that learn the noise profile on the fly. That's a genuine revolution for producers and engineers who need clean audio fast.

But not all noise reduction plugins are created equal. Some excel at broadband noise removal but smear transients. Others are surgical enough for forensic audio work but have a learning curve that makes them impractical for quick mix sessions. In this guide we've tested and evaluated every major contender across real-world use cases: home studio vocal recording, location audio for sync licensing, archival restoration, broadcast dialogue, and live performance. Here's everything you need to know to choose the right tool for your workflow.

How Noise Reduction Plugins Actually Work

Understanding the underlying technology helps you choose the right plugin and use it more effectively. Modern noise reduction plugins fall into three main processing paradigms, and many advanced tools combine all three.

Spectral Subtraction and Noise Profiling

The classical approach, still used in tools like iZotope RX's Voice De-noise and the core of many Waves algorithms. You capture a noise profile — a short recording of the noise alone, without the signal you want to keep — and the plugin analyzes its spectral fingerprint across hundreds of frequency bands. During processing, it continuously subtracts that fingerprint from the incoming audio. The result is effective against stationary noise sources (HVAC, electrical hum, consistent tape hiss) but struggles with variable noise like wind or crowd noise that changes over time.

Machine Learning and Neural Network Processing

The newer generation of noise reduction — exemplified by Waves Clarity Vx, NVIDIA RTX Voice, and iZotope RX 11's Machine Learning modules — uses deep neural networks trained on millions of audio samples to separate signal from noise without requiring a noise profile capture. These models learn what speech, singing, and musical instruments sound like and suppress everything that doesn't fit those learned patterns. The advantage is remarkable speed and effectiveness on non-stationary noise. The risk is over-processing artifacts when the model gets confused by unusual timbres or unconventional recording techniques.

Spectral Repair and Interpolation

Used in iZotope RX's Spectral Repair module and similar tools, this approach treats noise removal more like photo retouching. In the spectral view, the engineer manually selects noise regions and the algorithm fills them in using interpolated data from surrounding areas. This is the most time-intensive method but produces the most transparent results for isolated noise events — a car horn in the background of a vocal take, a cough during a string quartet recording, the click of a camera shutter in a live session.

Dynamic Noise Reduction

Some plugins, including Waves NS1 and the noise gate-based approaches in iZotope Neutron, use dynamic processing — essentially very sophisticated downward expansion — to suppress noise during quiet passages while leaving signal-dominant passages untouched. This is less transparent than true spectral processing but introduces far less latency, making it viable for tracking and live applications. It works best when your signal-to-noise ratio is already reasonably good and you simply need to tighten up the noise floor between phrases.

Top Noise Reduction Plugins: Quick Comparison

Best Noise Reduction Plugins: In-Depth Reviews

1. iZotope RX 11 Advanced — Best Overall

iZotope RX has been the industry standard for audio restoration and noise reduction since its early versions, and RX 11 Advanced represents the most comprehensive noise reduction toolkit available to anyone working in music production or post-production. It's not a plugin in the traditional sense — it operates as both a standalone application with a full spectral editing interface and as a suite of AAX, VST3, and AU plugins that connect back to the standalone app via RX Connect.

The flagship noise reduction module in RX 11 is Voice De-noise, which uses a machine learning model specifically trained for voice and vocal content. It distinguishes between speech, singing, and background noise with remarkable accuracy, and it does so in real time with a latency of around 20ms — acceptable for tracking sessions where you're monitoring through headphones and can dial in a small buffer. For post-production work, you can switch to the high-quality offline mode, which applies the same algorithm without any latency constraint and produces noticeably cleaner results on difficult material.

The Spectral De-noise module remains the best implementation of classical noise profiling available. You select a region of noise-only audio, click Learn, and the plugin builds a 4,096-band noise profile. The Threshold and Reduction sliders give you granular control over the trade-off between noise suppression and musical artifacts. The Artifact Control knob specifically addresses the common problem of processing artifacts — the metallic, watery quality that characterizes over-processed noise reduction. At aggressive settings you'll still hear artifacts, but they're considerably more musical than competing tools at similar reduction depths.

For music production specifically, the Dialogue De-reverb module (included in Advanced) is invaluable for dealing with roomy vocal recordings, and the Hum Removal module handles both 50Hz and 60Hz electrical hum with harmonic tracking up to the 10th partial. The De-wind, De-crackle, and De-clip modules round out a toolkit that handles essentially every noise problem you'll encounter in a recording career.

The price of RX 11 Advanced ($1,199 for a perpetual license) puts it out of reach for casual users, but for professional engineers it pays for itself quickly. The Standard edition at $399 covers the core noise reduction needs of most home studio producers — Voice De-noise, Spectral De-noise, Hum Removal, and De-click — and is the version we'd recommend for anyone who doesn't need the forensic capabilities of Advanced. Check our dedicated iZotope RX 11 review for a deeper teardown of every module.

Best for: Professional post-production engineers, film and TV audio, serious home studio operators who need the most transparent noise removal available.

Pros: Best spectral editor in the industry; ML + spectral hybrid approach; handles every noise type; massive module library in Advanced.

Cons: Expensive; steep learning curve for spectral editing; offline workflow adds steps to a live session.

2. Waves Clarity Vx Pro — Best for Real-Time Vocal Processing

Waves Clarity Vx Pro is the tool that's changed how a lot of working producers think about noise reduction in a mixing context. Launched initially as Clarity Vx, the Pro version extends the original with multi-stem processing — you can run separate noise reduction instances on voice, music, and effects layers, or use it in Stem mode to separate and clean multiple sources simultaneously in a single plugin instance.

The neural network at Clarity Vx Pro's core was trained on an enormous dataset of voice recordings across dozens of languages, accents, recording environments, and microphone types. The result is a plugin that works with essentially zero setup: drop it on a vocal track, set the Voice Sensitivity and Noise Reduction sliders, and you're done. In typical home studio conditions — modest room treatment, a decent large-diaphragm condenser like the AKG C414, moderate background noise — Clarity Vx Pro can remove 15-20dB of broadband noise while maintaining vocal naturalness that's genuinely impressive.

Where Clarity Vx Pro earns its "Pro" designation over the standard Clarity Vx is in the processing headroom it offers. The standard version provides a single reduction slider; the Pro adds a Transient and Tonal balance control, per-band processing visualization, and the ability to monitor the noise residual in isolation — a feature borrowed from RX's workflow that makes it much easier to dial in the right reduction depth without guessing. For producers working on vocal mixing who need noise reduction in the signal chain during mix-down rather than as a pre-processing step, this is the most practical tool available.

Latency is approximately 23ms in standard mode — almost imperceptible with hardware monitoring, perfectly workable with any normal mix session. The plugin runs entirely on your CPU without requiring a noise profile capture, meaning it adapts to changing noise environments in real time. This makes it uniquely useful for recording vocals in sessions where you can't control the environment — tracking in a living room, recording through a window, working with field recordings that have variable ambient noise.

The perpetual license sits at $199 (frequently on sale, as Waves plugins often are), and a Waves Creative Access subscription at $29/month gives you access to the entire Waves catalog including Clarity Vx Pro. If you're already using Waves plugins for mixing and mastering, this is likely the most cost-effective noise reduction upgrade you can make.

Best for: Producers and engineers who need noise reduction in the DAW signal chain without adding a separate offline processing step.

Pros: Genuinely real-time with low latency; no noise profile needed; natural-sounding results on vocals; multi-stem Pro mode.

Cons: Less effective on non-vocal content; can struggle with extreme noise conditions or unusual vocal timbres; subscription model adds ongoing cost.

3. Cedar DNS One — Best for Broadcast and Live Applications

Cedar Audio has been building noise reduction hardware and software for broadcast facilities and post-production houses since the 1990s, and the DNS One represents their flagship algorithmic approach distilled into a plugin format. Unlike the machine learning tools from iZotope and Waves, Cedar's DNS (Dialogue Noise Suppressor) uses an adaptive statistical algorithm — a sophisticated mathematical model of speech and noise that adapts in real time to changing noise conditions without requiring ML training data or a GPU.

This matters in professional broadcast contexts for several reasons. Cedar's algorithm introduces less than 1ms of processing latency, making it viable for live broadcast environments where any perceivable delay is unacceptable. It runs on standard CPU without any specialized hardware requirements. And it has been validated by major broadcast networks — BBC, NBC, CNN — for live on-air use, which means it has been stress-tested against the hardest possible noise conditions: reporters in the field, noisy press conferences, outdoor live locations.

The DNS One's interface is deliberately minimal: a Noise Suppression dial and a Learn button, plus a meter showing suppression depth. The simplicity is deceptive — the algorithm underneath is doing extraordinarily sophisticated work. At $995 it's expensive for a single-purpose plugin, but post-production facilities that work on broadcast content will find it earns its place quickly. For music production contexts, the DNS One is overkill unless you're doing a lot of session work on recorded dialogue or broadcast content.

Best for: Broadcast audio engineers, post houses handling live or near-live content, location sound professionals.

Pros: Sub-millisecond latency; broadcast-proven algorithm; handles highly variable noise superbly.

Cons: Expensive for its narrow use case; less versatile than RX for music production; no spectral editing.

4. Accusonus ERA Bundle Pro — Best Value for Beginners

The Accusonus ERA (Enhanced Repair Audio) Bundle has built a devoted following among podcasters, YouTube content creators, and home studio beginners who need good-enough noise reduction without any learning curve. The ERA Noise Remover Pro, the flagship noise reduction tool in the bundle, reduces to a single "Amount" knob that controls a sophisticated ML-based processing chain.

The single-knob philosophy is genuinely useful for workflows where you need to process a lot of clips quickly. Each knob position triggers a different internal processing preset, but the transitions between them are smooth enough that you can often dial in the right setting by ear without understanding the underlying algorithm. The ERA bundle also includes De-Reverb, De-Clip, De-Breath, Plosive Remover, and a Voice Leveler, making it a comprehensive voice cleanup toolkit for vocal production.

Results are good for moderate noise reduction — ERA handles steady-state noise (room tone, HVAC, USB mic noise floor) cleanly up to about 12-15dB of reduction before artifacts become noticeable. For heavier noise problems, the single-knob approach becomes a limitation because you can't tune the trade-off between reduction depth and artifact character the way you can in RX or Clarity Vx Pro. The subscription pricing at $199 per year gives you the full ERA Bundle, which represents solid value given the breadth of tools included.

Best for: Podcasters, content creators, home studio vocalists who want fast results without a technical learning curve.

Pros: Extremely easy to use; comprehensive bundle; good results on mild to moderate noise.

Cons: Limited control for heavy noise problems; subscription model; less transparent on extreme settings than RX or Clarity.

5. Waves NS1 Noise Suppressor — Best Budget Option

The Waves NS1 Noise Suppressor is a single-knob dynamic noise suppressor that's been a staple of broadcast and podcast workflows since its release. Like ERA, it reduces to a single "Amount" control, but the processing underneath is fundamentally different — NS1 uses adaptive downward expansion rather than spectral processing. It detects when signal is present and backs off, and detects noise-only passages and suppresses them aggressively.

This approach makes NS1 extremely effective for its primary use case: reducing background noise in dialogue, podcast vocals, and broadcast voice content where the signal-to-noise ratio is already decent. It's particularly transparent on voice content because the algorithm doesn't touch the spectral character of the signal at all — it only adjusts level dynamically. The cost, typically around $29 on sale (Waves runs frequent promotions), makes it one of the most accessible noise suppression tools available.

The limitation of the dynamic approach is that it can't address noise that co-exists with signal — if your noise floor is clearly audible during passages where the vocalist is singing, NS1 can't cleanly remove it without affecting the vocal itself. For those situations you need spectral processing. But for the common problem of noisy room tone between phrases, NS1 does the job cleanly and cheaply.

Best for: Podcasters, broadcasters, producers who need a quick noise gate-style suppressor for voice content.

Pros: Very affordable; extremely simple; transparent on voice; low latency.

Cons: Dynamic approach can't clean noise under signal; single-knob limits control.

6. Bertom Denoiser Pro — Best Value Spectral Plugin

Bertom Audio has built a reputation for delivering high-quality, fairly priced plugins, and the Denoiser Pro is the best budget spectral noise reduction plugin available in 2026. At $49 it's within reach of any home studio operator, and the quality of the spectral processing punches well above its price point.

The workflow mirrors the classical iZotope approach: capture a noise profile, set reduction depth and artifact control, monitor residual. The UI is clear and functional, the spectral analysis is detailed enough for serious work, and the processing adds minimal coloration to the signal. Bertom Denoiser Pro handles broadband noise, electrical hum harmonics (with a separate HF/LF tilt control for the noise profile), and can process at up to 192kHz for high-resolution work.

The plugin doesn't have the ML capabilities of RX or Clarity Vx, which means it requires a noise profile capture and will struggle more than those tools on non-stationary noise. But for the most common home studio scenario — consistent room noise from an HVAC system, a consistent noise floor from your mic preamp chain — Bertom Denoiser Pro delivers results that are genuinely competitive with tools costing five to ten times more. If you're on a tight budget and need proper spectral noise reduction rather than a simple noise gate, this is the plugin to get.

Best for: Budget-conscious home studio producers who need spectral noise profiling without paying iZotope prices.

Pros: Affordable; genuine spectral processing; supports high sample rates; good artifact control.

Cons: No ML processing; requires noise profile; less comprehensive than RX.

7. NVIDIA RTX Voice / Broadcast — Best Free Option

NVIDIA's RTX Broadcast (which incorporates RTX Voice's noise removal capabilities) is technically not a DAW plugin — it operates as a virtual audio driver that intercepts microphone input before it reaches your DAW. But it's so effective and so widely used among streaming producers and remote recording artists that it deserves a place in this guide.

Running on the Tensor Cores of an NVIDIA RTX GPU, the noise removal model processes audio in real time with near-zero perceivable latency. In controlled tests, it removes keyboard noise, fan noise, HVAC rumble, and even moderate crowd noise more aggressively than any CPU-based real-time plugin, without the artifacts that would occur at equivalent reduction depths in a CPU-based spectral processor. The reason is raw compute power: the GPU can run a significantly larger neural network model than any CPU plugin can afford to run in real time.

The obvious limitation is hardware dependency — you need an NVIDIA RTX 2070 or better to run it, and it's not available as a traditional DAW plugin. For producers who already own qualifying hardware, it's completely free via NVIDIA Broadcast 1.4 or later. For those without compatible hardware, the cost of an RTX GPU puts it in a different economic category. Worth mentioning also that AMD has released AMF Noise Suppression with similar GPU-accelerated capabilities for Radeon RX 6000 series and newer cards.

Best for: Streamers, remote session musicians, producers who record while streaming or doing video calls alongside their DAW sessions.

Pros: Free for compatible hardware; extremely effective ML noise removal; near-zero latency.

Cons: Requires NVIDIA RTX GPU; not a DAW plugin; no fine control over processing.

How to Choose the Right Noise Reduction Plugin

With seven credible tools in the market across a wide price range, choosing the right one comes down to clarifying your primary use case, your workflow requirements, and your budget. Here's a framework for making that decision systematically.

Identify Your Primary Noise Problem

Not all noise is the same, and different plugins address different noise types with different effectiveness. Before spending money, identify which of these categories describes your most common problem:

Consistent broadband noise (room tone, mic noise floor, tape hiss): Any spectral denoiser handles this well. Bertom Denoiser Pro at $49 is sufficient. iZotope RX Standard at $399 is better. RX Advanced is overkill.

Variable ambient noise (traffic, people, wind, varying HVAC): Machine learning tools significantly outperform classical spectral subtraction here. Waves Clarity Vx Pro or iZotope RX Voice De-noise are your best options.

Electrical hum and buzz (50/60Hz and harmonics): iZotope RX Hum Removal is the gold standard. Many EQs with precise notch filtering can handle mild hum — check our best EQ plugins guide for options that include surgical notch capabilities.

Click and crackle (vinyl transfers, digital glitches, microphone handling noise): iZotope RX De-click and De-crackle, or the click removal module in Izotope RX. No other tools match RX here.

Reverb and room sound (overly roomy recordings): iZotope RX De-reverb (Advanced only), or plug-in solutions like Zynaptiq UNFILTER. This is distinct from noise removal but often part of the same restoration workflow.

Assess Your Workflow Requirements

The distinction between offline and real-time processing is one of the most important workflow factors. If you're doing post-production work — cleaning up recorded material that will be edited later — offline processing in iZotope RX's standalone application gives you the best quality because the algorithm can look ahead and behind in time without any latency constraint. The results are noticeably more transparent than real-time processing, especially on difficult material.

If you're recording or mixing in real time — monitoring through your DAW during a vocal session, processing a live broadcast feed, running a podcast recording — you need a plugin with low enough latency to stay inside your monitoring chain. Waves Clarity Vx Pro at ~23ms, Cedar DNS One at under 1ms, and Waves NS1 (essentially zero latency) are the tools built for this context. For more on building efficient processing chains in the DAW, see our guide to how to build a plugin chain.

Budget Allocation

If you're just starting out and need noise reduction as part of a broader plugin investment, consider that the ERA Bundle Pro at $199 per year gives you noise removal plus de-reverb, de-breath, and plosive removal — a comprehensive voice cleanup toolkit. For the same annual cost you could spend $199 on a Waves Creative Access subscription and get Clarity Vx Pro plus hundreds of other mixing plugins.

For producers who are investing in their toolkit more seriously, iZotope RX 11 Standard at $399 is the best single investment in audio quality you can make. Even if you only use Voice De-noise and Spectral De-noise, the improvement in vocal and recording quality it enables will be audible on every project you work on from now on.

Noise Reduction in the Music Production Workflow

While noise reduction is most commonly associated with post-production and broadcast, it has equally important applications in music production. Understanding where and when to apply noise reduction in a music production workflow can dramatically improve the transparency of your mixes.

Pre-Processing vs. In-Mix Processing

The most important workflow decision in music production is whether to apply noise reduction as an offline pre-processing step before loading audio into your DAW session, or as a real-time plugin in your mixing chain. Both approaches have valid use cases.

Pre-processing with iZotope RX is the right choice when you have a significant noise problem that needs substantial reduction — more than about 10-12dB. Processing offline lets you use the highest-quality algorithm modes, review the spectral view to confirm you're not damaging the signal, and commit a cleaned file to disk before mixing. This keeps your mix session CPU-efficient (no heavy processing plugin running) and lets you make permanent, inspectable decisions about the audio quality.

In-mix noise reduction with Clarity Vx Pro, NS1, or ERA Noise Remover is the right choice for mild noise problems where you want to preserve the option to adjust or bypass the processing later. It's also useful in situations where you're unsure about the right reduction depth and want to experiment during mixing. The main risk is that aggressive in-mix noise reduction can interact with other processing — particularly compression and saturation — in ways that make artifacts more audible. Always put your noise reduction plugin first in the chain, before any dynamic processing, so the compressor is responding to clean signal rather than amplifying noise residual.

Specific Music Production Applications

Acoustic guitar recordings: Acoustic guitar is particularly susceptible to room noise because the dynamic range between picking attacks and decay tails is wide — noise becomes clearly audible during the quiet sustain and decay of notes. A noise profile captured during a pause between takes, applied with moderate reduction in Spectral De-noise, cleans up the decay tails dramatically without affecting the attack transients. See our guide on how to record acoustic guitar for microphone placement strategies that minimize the noise problem in the first place.

Vintage sample processing: Lo-fi production styles often incorporate vinyl crackle and tape noise intentionally, but for more pristine uses of vintage samples, iZotope RX's De-crackle and De-noise combination is standard workflow. Start with the minimum effective De-crackle setting — just enough to reduce the worst clicks — then apply gentle broadband noise reduction to tame the noise floor. Over-processing kills the vintage character that made the sample worth using.

Live recording clean-up: Live recordings captured in imperfect environments — rehearsal rooms, stages with poor isolation, outdoor venues — benefit enormously from RX processing before mixing. Crowd noise between songs, PA bleed into microphones, and variable room rumble can all be addressed with a combination of Voice De-noise (for vocal microphones), Spectral De-noise (for instrument mics), and dynamic processing to clean up the quieter passages. This workflow is extensively documented in our iZotope RX workflow guide.

Home studio vocal recording: The single most impactful use of noise reduction for most home studio producers. A bedroom recording with a condenser mic will almost always have room noise, computer fan noise, and potentially HVAC noise in the background. Processing the vocal with Waves Clarity Vx Pro or iZotope RX Voice De-noise before mixing brings the vocal into a professional clarity range that would otherwise require either a treated studio or extensive manual editing. Pair this with proper gain staging — check our guide on how to record vocals in a home studio for the complete workflow.

Avoiding Common Noise Reduction Mistakes

Over-processing is the most common mistake. Noise reduction artifacts — the watery, metallic, or hollow quality of over-reduced audio — are often more distracting than the original noise. Train your ears to hear these artifacts by listening to the noise residual in isolation (most professional tools let you solo the removed signal). If you can hear fragments of the musical signal in the residual, you've gone too far and need to back off the reduction depth.

The second most common mistake is applying noise reduction too late in the signal chain. Compression amplifies everything, including noise. Saturation and harmonic excitation can interact with the noise floor in unpleasant ways. And reverb applied to a noisy signal will make the reverb tail obviously noisy. Always position noise reduction first in the processing chain, ideally as a pre-processing step before any other treatment.

Finally, don't neglect the source. Noise reduction is a rescue tool, not a substitute for good recording practice. Acoustic treatment in your recording space — even basic panel absorbers and bass traps — will reduce the problem before it needs to be solved. A quality audio interface with a clean preamp will lower the noise floor at the source. These upstream improvements compound — every 3dB improvement in signal-to-noise ratio at the source means you need 3dB less noise reduction, which means 3dB less risk of processing artifacts. Browse our home studio acoustic treatment guide for practical, budget-friendly room treatment strategies.

Advanced Noise Reduction Techniques

For producers and engineers ready to go beyond single-click noise reduction, these advanced techniques unlock better results on difficult material.

Multi-Stage Noise Reduction

As mentioned in the highlight box above, the most transparent results on heavily noisy material come from stacking multiple tools at mild settings rather than pushing one tool to its limits. A typical multi-stage chain might look like this:

Stage 1: iZotope RX De-hum to remove 60Hz fundamental and harmonics at maximum precision. Stage 2: RX Spectral De-noise at 8-10dB reduction to address the broadband noise floor. Stage 3: Waves Clarity Vx Pro at 20-30% to catch any remaining variable noise during signal passages. Stage 4: Waves NS1 at moderate setting to suppress residual noise in the quietest passages. Each stage operates in its area of strength at a conservative setting, and the cumulative effect is a degree of noise reduction that would introduce severe artifacts if attempted with a single tool.

Spectral Editing in iZotope RX

The spectral editor in iZotope RX — available in Standard and Advanced — is the most powerful individual tool in the noise reduction toolkit. The display shows audio as a spectrogram with time on the horizontal axis and frequency on the vertical axis, with color representing amplitude. Noise appears as a consistent texture across the spectrogram; transient events (clicks, coughs, car horns) appear as vertical lines or bright spots.

The Spectral Repair tool lets you select any region in the spectrogram and replace it with interpolated content from the surrounding material. For isolated noise events — a single loud click, a phone notification sound in the background of a vocal take, a brief burst of electrical interference — this produces completely invisible results that no automated algorithm can match. The learning curve is real, but once you've spent a few hours in the spectral view, you'll find uses for it on almost every complex audio repair job.

Noise Reduction for Stereo and Surround Content

For stereo material, always process both channels simultaneously with linked settings rather than treating left and right independently. Independent channel processing can introduce subtle phase shifts and spatial imaging artifacts that make the stereo image feel unstable or narrow. iZotope RX handles stereo processing correctly in all its modules, but check that third-party tools are operating in true stereo or mid-side mode rather than dual-mono when processing stereo files.

For surround and immersive content — increasingly relevant for producers working in Dolby Atmos — the same principle applies, extended across all channels. iZotope RX Advanced supports surround processing natively. When applying noise reduction to Atmos content, process the bed channels and object audio separately, as they typically have different noise characteristics. For more on immersive audio workflows, see our guide on how to mix in Dolby Atmos.

Machine Learning Model Selection and Sensitivity Tuning

In tools that offer model selection — including iZotope RX 11's ML modules and Waves Clarity Vx Pro — choosing the right model for your content type can make a significant difference. RX 11 Voice De-noise includes models trained specifically for speech, singing, and broadband content. Using the singing model on spoken word dialogue, or vice versa, produces subtly worse results than matching the model to the content. Take thirty seconds to identify the right model before processing.

Sensitivity settings (called Threshold, Amount, or Sensitivity depending on the tool) should always be set at the minimum effective level. Find the point where the noise reduction is just sufficient to solve the problem, not the maximum point where the noise disappears entirely. Especially with ML-based tools, the difference between "noise is inaudible" and "noise is gone" often corresponds to the difference between transparent processing and audible artifacts.

Practical Exercises

Frequently Asked Questions