Is iZotope RX worth it for cleaning up voice recordings?
RX 12 Elements costs $99 and Standard $399, but the real split is the editor and Dialogue Isolate. Which modules a voice actually needs, what old RX 7 licences are worth, and when fixing the input beats repairing every file.
Twelve minutes into a forty-minute interview, the fridge behind you kicks on and hums under every answer until the end. Your guest lives in another time zone, so a retake is not happening. That is the exact situation iZotope RX was built for: a recording that already exists, cannot be made again, and has something baked into it that should not be there.
It is also the situation most people searching "izotope rx" are not actually in. Plenty of them have a mic that sounds bad every day, on every call and every take, and they are hoping a repair suite will fix that. It can, one file at a time, forever. Whether RX is worth the money depends almost entirely on which of those two problems you have.
If you only need one noisy file cleaned once and would rather not buy anything, start with how to remove background noise from a voice recording, which walks through the free tools first. This page is for the reader deciding whether RX's price buys something those free tools cannot.
The short answer
RX is worth it when you repair spoken audio regularly and the recordings cannot be redone: interviews, field recordings, lectures, client files, dialogue for video. That is the job RX is designed around, and the two editions most voice people choose between cost $99 and $399 as of October 2026.
RX is a poor buy when the real problem is the sound going into your mic every day. Repairing every file after the fact is a tax you pay on each recording. Fixing the input once is cheaper, and on a live call it is the only option, because there is no file to repair.
The rest of this article covers what you get in each edition, which modules matter for voice, a working order for a spoken-word cleanup, and where repair stops being the right tool.
What iZotope RX is in 2026: RX 12 and three editions
The current version is RX 12. iZotope's store sells three editions, plus a bundle and two subscriptions that include it. Prices below are list prices from iZotope's own pages as of October 2026; during a sale the checkout price can be lower.
| Edition | Price (USD) | What you get | Standalone editor | Voice tools that matter |
|---|---|---|---|---|
| RX 12 Elements | $99 | 6 plugins for a host app | No | Voice De-noise, De-hum, De-click, De-clip, De-reverb, Repair Assistant |
| RX 12 Standard | $399 | Editor + 18 plugins | Yes | Everything above + Dialogue Isolate, Mouth De-click, De-ess, De-plosive, Breath Control, Spectral De-noise, Spectral Repair, Loudness Control |
| RX 12 Advanced | $1,399 | Editor + 50+ tools | Yes | Everything above + Dialogue Contour, De-wind, De-rustle, EQ Match, Leveler, Ambience Match |
| iZotope Plus subscription | $15/month or $150/year | Includes RX 12 Elements | No | Same as Elements |
| iZotope Pro subscription | $25/month or $250/year | Includes RX 12 Standard | Yes | Same as Standard |
The subscriptions come with a free 7-day trial and are run through Native Instruments. If you already own an iZotope product, the Standard page says you may qualify for a lower upgrade price once you log in, which matters if you are sitting on an old licence.
RX 12 is confirmed by iZotope on macOS Sonoma 14.7, Sequoia 15.7 and Tahoe 26.2, on both Apple silicon and Intel Macs, and on Windows 10 and 11. Plugin formats are AU, AAX, VST3 and ARA variants, all 64-bit. VST2 is not on the list, so a host that only loads VST2 will not see it.
RX 7, RX 8, RX 9: what about the older versions?
Search data shows "izotope rx 7" alone still gets more monthly searches than RX 8 and RX 9 combined. Much of that traffic is people following an old iZotope RX tutorial or reading an RX 7 review from years ago; the rest is mostly people shopping for used licences, all the way back to RX 5 Advanced.
Three things to know before you chase an old version:
- iZotope's store lists RX 12 editions only. Its own version comparison table goes back to RX 9, so anything older is outside what the vendor documents on the product page.
- Module names have carried over well. Voice De-noise, De-hum, De-click, De-reverb and Spectral Repair exist in RX 12, so an RX 7 tutorial still teaches the idea. The layout has moved: RX 12 added a module search bar, a larger default spectrogram and a resizable History panel, so menu paths in older videos will not match.
- iZotope confirms compatibility only for the current version's list of macOS releases. If you plan to run an older RX on a new Mac, check iZotope's support pages for that version before paying for a second-hand licence. A cheap licence that will not run on your system costs more than the upgrade price.
If you already own RX 7 or RX 8 and it still runs, keep using it for steady noise and hum. What you miss is mostly the newer machine-learning modules, especially the improved Dialogue Isolate.
The modules that matter for voice (and the many that do not)
RX Advanced has more than fifty tools. A podcaster, teacher or video maker needs about a dozen of them. Here they are, matched to the problem you actually hear.
| What you hear | Module | Lowest edition |
|---|---|---|
| Steady hiss, air conditioning, fan, street wash | Voice De-noise (Adaptive mode follows a changing noise floor) | Elements |
| Low buzz from mains power or a ground loop | De-hum | Elements |
| Sharp clicks and digital pops | De-click | Elements |
| Distortion on loud words | De-clip | Elements |
| Kitchen or classroom echo | De-reverb | Elements |
| Wet mouth sounds, lip smacks | Mouth De-click | Standard |
| Speech buried under changing noise or reverb | Dialogue Isolate | Standard |
| Harsh "s" sounds | De-ess | Standard |
| Thumps on "p" and "b" | De-plosive | Standard |
| Loud breaths between phrases | Breath Control | Standard |
| A dog bark or siren over one word | Spectral Repair in the editor | Standard |
| Clothing rustle on a clip-on mic | De-rustle | Advanced |
What a voice person can ignore: Music Rebalance, Scene Rebalance, Guitar De-noise, Azimuth, Wow & Flutter and the stem tools. They are built for music, film and restoration work. Some of them, such as Scene Rebalance, Azimuth and Wow & Flutter, exist only in Advanced, which is part of why it costs $1,399.
Dialogue Isolate, the module people buy Standard for
People search for "izotope rx dialogue isolate" by name for a reason. Dialogue Isolate is a machine-learning module that separates speech from everything around it, which is a different job from Voice De-noise. A de-noiser learns a noise floor and subtracts it, so it works best on noise that stays roughly the same. Dialogue Isolate tries to keep the voice and drop everything else, and iZotope pitches it at speech buried under noise and reverb. In principle that lets it reach noise that does not hold still, which a classic de-noiser struggles with.
It is in Standard and Advanced, not Elements. iZotope says RX 12 improved its neural nets in both offline and real-time modes, and Advanced adds a high-quality offline mode and multi-band processing. If your recordings come from noisy, changing places, this one module is most of the case for Standard over Elements.
De-hum and the 50 versus 60 Hz question
Mains hum sits on the power-line frequency and its multiples: 50 Hz in Europe, Russia and most of the world, 60 Hz in the US, Canada and much of the Americas. De-hum in Static mode targets a fundamental plus 16 harmonic bands above it; Dynamic mode can learn and remove hum with up to 1,024 variable notch filters. Set the wrong base frequency and you notch out the wrong part of the voice while the buzz stays put, so check where the file was recorded.
Elements vs Standard: the real difference is the editor
Most comparisons list modules. The bigger difference is that RX 12 Elements has no standalone editor.
Elements is six plugins. They run inside a host: iZotope lists Logic Pro 12, Pro Tools 2025, Ableton Live 12, Reaper 7, Adobe Audition 2026, Premiere Pro 2026 and DaVinci Resolve 19 among confirmed hosts. GarageBand is not on that list, so if GarageBand is all you use, check compatibility before you pay. Inside the host, you drop Voice De-noise or De-hum on a track and adjust it like any other effect.
Standard adds the RX Audio Editor: a separate app with a spectrogram, where you see sound as a picture of frequency over time and can select a single cough, a phone buzz or a chair squeak and repair just that region. It also brings the Batch Processor, Module Chain, Loudness Control and Loudness Optimize, and the modules from the table above.
A practical split:
- Elements if your problems are steady: one kind of hiss, one hum, a little room echo, and you already work in one of the listed hosts.
- Standard if your problems are events: a cough over a key sentence, a siren, a mouth click on every third word, speech under café noise. The editor plus Dialogue Isolate and Mouth De-click are the reason, and they do not exist in Elements.
- Advanced if you deliver dialogue for film or TV and need Dialogue Contour, EQ Match, Ambience Match or De-wind. For podcasts, lessons and YouTube voice-overs, the jump from $399 to $1,399 buys tools most people never open.
If you will use RX for a few months and stop, the $25 Pro subscription with Standard inside is cheaper than buying. If you will use it for years, $399 once beats $250 a year by the second year.
A starting order for cleaning one spoken-word file
There is no universal preset, but the order matters more than any single setting, because each stage hands its output to the next. A starting point for a voice file in RX Standard:
Stage Module Starting point
1 De-hum 50 Hz (Europe, Russia) or 60 Hz (US, Canada)
2 Mouth De-click only if you hear lip smacks
3 Voice De-noise Adaptive mode only if the noise floor changes; Dialogue Isolate for noise that moves
4 De-reverb lightest setting that takes the edge off the room
5 De-ess only if the s sounds now poke out
6 Loudness Control podcast target: about -16 LUFS, true peak no higher than -1 dB
The logic behind the order:
- Hum first. Hum is tonal and predictable, so it is the easiest thing to remove cleanly. Taking it out first means the de-noiser later does not spend its effort on it.
- Clicks before broad noise reduction. Clicks are short events; deal with them while they are still distinct.
- Noise reduction in the middle. This is where most artifacts come from, so use the least that works. If you start from Repair Assistant, it offers light, medium and aggressive versions of each suggestion. Start with light.
- De-reverb after the noise. Room echo and noise overlap; working on a cleaner signal is easier.
- De-ess late, because noise reduction and de-reverb can change how bright the voice sounds.
- Loudness last. The -16 LUFS and -1 dB true-peak figures come from Apple's podcast audio requirements; other platforms use different targets, so check yours.
Listen to the removed part, not only the result. If the module you are using can play back only what it takes out, use that: if you can hear words in the removed signal, the setting is eating your voice along with the noise.
Where RX stops: repair after the fact vs a clean input
Every repair module makes a guess. It estimates what the voice under the noise was and rebuilds that estimate. With light noise the guess is good. With heavy noise, the module has to decide what belongs to the voice, and pushed hard it takes some of the voice away: a watery, metallic or lisping edge on consonants. That is not a flaw specific to RX; it is the nature of separating two sounds that overlap in frequency and time.
So the most effective use of a $399 repair suite is to need it less. A few input changes do more than any slider:
- Distance. Sound level drops about 6 dB each time distance doubles. Moving a mic from 40 cm to 10 cm from your mouth raises your voice by about 12 dB against the room, before any software touches it.
- The room. Soft furniture and a closed door reduce the echo De-reverb would otherwise have to remove.
- The noise source. Switching off the fan for twenty minutes costs nothing. RX's best output on a fan recording is still a guess.
Live calls are a different problem entirely
RX works on recordings. On a Zoom or Google Meet call there is no file, only a live stream of audio leaving your Mac. RX plugins can run in real time inside a host, and Voice De-noise is described by iZotope as zero-latency, but getting that processed sound into a call app means building a chain: a plugin host, plus a virtual audio device to carry the output into Zoom. Audio Hijack and Loopback covers what that routing involves. It works, and it is a lot of moving parts for a daily meeting.
The simpler built-in options come first. On macOS Sonoma or later, the Mic Mode menu offers Voice Isolation in apps that support it. Zoom has its own background noise suppression with Auto, Low, Medium and High levels. Both are free and worth trying before you buy anything.
This is the gap TunedMic is built for. TunedMic adds a virtual microphone to your Mac that any app can choose, the same way you pick a USB mic. A one-minute Sound Check measures your voice, mic and room, then TunedMic processes the sound live on the Mac with less than a tenth of a second of delay, and every app that picks it hears the processed voice: Zoom, Meet, Teams, Discord, OBS, FaceTime. Recordings made through it start cleaner, so there is less left for RX or anything else to repair.
Its limits are plain. TunedMic does not open existing files, so it cannot rescue the fridge interview from the first paragraph; for that you need a repair tool. It needs a Mac with Apple silicon (M1 or newer) and macOS 14 or later, and there is no Windows version. It will not remove every noise or turn a cheap mic into a studio one.
Cheaper routes to the same result
Before spending $99 or $399, run your worst file through the free options. If they get close enough, keep your money.
- Audacity (version 4.0.1, free) has Noise Reduction that works from a noise profile. It handles steady noise such as hiss and hum, not one-off sounds like a door slam.
- Adobe Podcast Enhance Speech has a free web tier for audio-only files up to 30 minutes, with a daily limit of one hour and no strength control. It is fully automatic, which also means you cannot dial it back.
- Voice Memos on the Mac has Enhance Recording, which reduces background noise and room echo on a file already recorded.
- Adobe Audition sits between free and RX: noise reduction, a spectral editor and loudness matching for US$22.99 a month on the annual plan billed monthly, as of October 2026. If you also want a multitrack editor, read whether Adobe Audition is worth it for voice cleanup.
If what you want is a one-click "make it sound better" tool, and you are not sure whether you need a file processor or something that runs live, the guide to audio enhancers separates the two kinds.
So, is iZotope RX worth it for you?
| You are | Buy | Why |
|---|---|---|
| Podcaster editing interviews you cannot re-record | RX 12 Standard, or the Pro subscription first | Dialogue Isolate, Mouth De-click, spectral repair of single events |
| Video maker with steady fan or AC noise, already in Resolve or Premiere | RX 12 Elements | Voice De-noise and De-hum inside the editor you use |
| Film or TV dialogue editor | RX 12 Advanced | Dialogue Contour, EQ Match, Ambience Match, De-wind |
| Someone who needs one file fixed once | Nothing yet | Audacity or Enhance Speech first |
| Someone whose mic sounds bad on every call | Not RX | Fix the input: placement, Voice Isolation, a live processor |
| Owner of RX 7 or 8 that still runs | Keep it, check upgrade pricing | Steady noise tools have not changed much |
RX earns its price on recordings that matter and cannot be made again. It is the wrong tool for a mic that sounds bad every single day.
Your next step
Pick your worst existing recording and run it through Audacity's Noise Reduction tonight. If the result is close enough, you are done for free. If it is not, start the 7-day iZotope subscription trial with the Pro plan, because it includes RX 12 Standard and lets you test Dialogue Isolate on that same file before paying $399.
If the problem is your sound on calls and new recordings rather than one old file, watch the 38-second video on the TunedMic home page: it shows the one-minute read-aloud that tunes the mic to your voice, and the Speech Report you get after a call. The free trial runs 7 days from first launch with every feature and no card or account. Payments are final and there are no refunds, so spend those seven days on your own calls. After that, it is $19 once on the pricing page.