> ## Content Index > Fetch the complete content index at: https://izotoperx.net/llms.txt > Use this file to discover other available public pages before exploring further. # What's New in iZotope RX 12: Every Major Change - URL: https://izotoperx.net/izotope-rx-12-new-features/ - Published: 2026-09-04T09:28:32.000Z - Updated: 2026-09-07T21:31:01.000Z - Description: RX 12 adds three headline workflows, rebuilds two familiar modules, updates its separation models and removes friction from the editor. Here is what each change actually does. - Author: Brandon Hayes - Tags: Getting Started, RX 12, Workflow **iZotope RX 12 adds three headline features: Scene Rebalance, Stems View and Trim Silence.** It also rebuilds De-bleed and Breath Control with machine learning, improves the neural models in Music Rebalance, Dialogue Isolate and Repair Assistant, and updates the editor with a larger spectrogram, module search and a resizable history panel. RX 12.0.0 was released on April 28, 2026. The edition matters. Scene Rebalance and Trim Silence require Advanced. Stems View, the rebuilt De-bleed and Breath Control, and the Music Rebalance/Dialogue Isolate improvements are in Standard and Advanced. Elements does not get those modules, but it does receive the improved Repair Assistant model. This guide inventories what changed and where to go next; it does not turn the feature list into an automatic upgrade recommendation. ## RX 12 new features at a glance | Change | Type | Edition | What it changes | | ---------------- | -------------------------- | ---------------------------- | ------------------------------------------------------------------------------------------- | | Scene Rebalance | New module | Advanced | Separates dialogue, music and effects from a stereo program mix | | Stems View | New workflow | Standard, Advanced | Opens isolated components as tracks for stem-specific RX repair | | Trim Silence | New module | Advanced | Detects and standardizes long-file gaps | | De-bleed | Rebuilt module | Standard, Advanced | Reference-free ML mode and real-time plug-in; Classic reference mode retained in the editor | | Breath Control | Rebuilt module | Standard, Advanced | New ML detection for speech and vocal breaths | | Music Rebalance | Improved model and plug-in | Standard, Advanced | Updated stem separation plus a real-time plug-in | | Dialogue Isolate | Improved model | Standard, Advanced | Updated separation of speech, reverb and noise | | Repair Assistant | Improved model | Elements, Standard, Advanced | Updated assistive repair analysis | | Editor design | Workflow update | Standard, Advanced app | Larger spectrogram, module search and resizable history | The categories follow the [official RX 12 upgrade overview](https://www.izotope.com/community/blog/rx-12?ref=izotoperx.net) and current edition matrix. “New” means RX 11 did not have that module or workflow. “Rebuilt” means the task existed but its engine or operating model changed. “Improved” means iZotope updated a neural model or added a plug-in route. A larger interface is useful, but it is not a restoration algorithm. ## RX 12 release date and current version As checked on September 7, 2026 UTC, the [RX Standard release notes](https://www.izotope.com/pages/release-notes/rx-standard?ref=izotoperx.net) list 12.0.0 as the latest RX 12 entry and date its initial major release to April 28, 2026\. The same notes give the current operating-system, host and plug-in-format support. Check them before installing because a feature can be relevant while the required host configuration is not. RX 12 is 64-bit only. The listed formats include AU, AAX, AAX AudioSuite, VST3 and selected ARA routes. Scene Rebalance, Ambience Match and De-rustle are AudioSuite-only as plug-ins. Spectral Editor AU ARA in Logic still requires Rosetta on Apple-silicon Macs. Those limits are not headline features, but they change how you can use the new tools. For a complete product verdict, use our [RX 12 review](https://izotoperx.net/izotope-rx-12-review/). If the question is whether to pay to move from RX 11, the narrower [RX 12 versus RX 11 upgrade guide](https://izotoperx.net/izotope-rx-12-vs-rx-11/) compares the changes with the work RX 11 already covers. ## Scene Rebalance separates dialogue, music and effects Scene Rebalance adds a new way to rebalance finished soundtracks. It takes a single stereo render and separates it into dialogue, music and effects components. You can change their relative levels when the original session or delivery stems are unavailable. A documentary scene with score masking an interview, a trailer with dialogue buried under effects, or a social-video export with the wrong balance are the intended class of problems. The [current Scene Rebalance manual](https://docs.izotope.com/rx12/en/scene-rebalance.html?ref=izotoperx.net) documents the three gains and the stem-splitting route. The result can feed Stems View for deeper work on one component. Scene Rebalance is included only in Advanced, and its plug-in form is AudioSuite. That AudioSuite plug-in has no pre-render Preview, so render to a protected copy and compare it with the original. Stem Split ignores the component Gain sliders and uses 0 dB gains; a split does not bake in the balance you just auditioned in the editor. Separation is not the same as receiving clean original stems. Dialogue and music often overlap in pitch, time and ambience. Test for watery speech, effects pulled into the dialogue stem, music transients that smear, and background tone that changes when a slider moves. A modest level adjustment may be more believable than trying to reconstruct a completely different mix; stop where leakage becomes more distracting than the original imbalance. ## Stems View turns separation into an editing workspace Stems View is the workflow change that connects separation with the rest of RX. The official [RX 12 feature inventory](https://www.izotope.com/products/rx-advanced?tab=features&ref=izotoperx.net) shows three entry points: Scene Rebalance, Music Rebalance and Dialogue Isolate. After splitting, the components open in a dedicated Stems tab. The main editor now also provides a Stem Split pane: choose an available splitter and click Process. In All Stems view, active lanes determine what plays and what receives edits; selecting one component in the header gives it an ordinary single-stem editing view. That changes the job from “turn the vocal down” to “repair the problem inside the vocal stem.” A hum that mainly follows bass can receive De-hum without processing drums. In Advanced, a dialogue stem can receive De-rustle while music and effects remain untouched. Stems View does not unlock modules missing from your edition. Separation artifacts still exist inside the stem, so each subsequent repair must be judged against the recombined result. Stems View is available in Standard and Advanced, but the splitter determines what you can create. Standard has Music Rebalance and Dialogue Isolate routes; Advanced also has Scene Rebalance. Our dedicated [Stems View guide](https://izotoperx.net/izotope-rx-stems-view/) covers splitting, repairing, recombining and export decisions. The [Stem Split pane manual](https://docs.izotope.com/rx12/en/stem-split-pane.html?ref=izotoperx.net) makes two limits explicit: Compare is unavailable in All Stems view, and File > Export there creates separate stem files. Use Export as Single File for the sum. Switch to an individual stem when you need Compare, then return to the full mix for approval. ## Trim Silence standardizes long-file gaps Trim Silence is new in Advanced. It detects passages below a threshold and lets editors shorten those gaps, but it ignores the current selection and always processes the entire file. Test on a separate copy. Do not independently trim synchronized microphones, stems or audio locked to picture: removing time changes their alignment. The current [Trim Silence manual](https://docs.izotope.com/rx12/en/trim-silence.html?ref=izotoperx.net) documents Threshold, Post-roll, Silence and Crossfade. Post-roll keeps audio after a silent segment begins; Silence sets the length to which longer gaps are trimmed after that post-roll, and Crossfade smooths the removal boundary. These controls still need listening checks. Its value rises with duration: a six-hour audiobook, a series of interviews or a long dialogue assembly contains far more repetitive gap work than a three-minute song. The danger is confusing low level with disposable silence. Breaths, quiet acknowledgements, dramatic pauses, room-tone bridges and sentence tails may all fall below a crude threshold. Test the quietest speaker, leave post-roll, audition every edit boundary and keep a reversible version. If the mistakes take longer to review than manual editing would have taken, the threshold is not ready. The [RX podcast cleanup workflow](https://izotoperx.net/izotope-rx-podcast-cleanup/) places Trim Silence before detailed polish only when the source is consistent enough to supervise safely. It is automation for repetitive structure, not permission to erase pacing. ## De-bleed adds reference-free repair and keeps Classic De-bleed existed before RX 12, so it belongs in the rebuilt column. The new machine-learning mode requires no separate reference track, and RX 12 adds a real-time plug-in using that approach. The Audio Editor also retains Classic, the RX 11 reference-based algorithm. Classic is not available in the new plug-in. The reference-free route removes alignment and routing friction from problems such as headphone spill into a vocal microphone or one instrument bleeding into another mic. The [RX 12 De-bleed manual](https://docs.izotope.com/rx12/en/de-bleed.html?ref=izotoperx.net) documents both approaches. In the new mode, Signal Type identifies the wanted source to isolate, not the unwanted bleed; choose a category that actually fits the recording. In Classic, the bleed source and active file must share a sample rate and be aligned within a few milliseconds before learning their relationship. Real-time audition makes it easier to hear the repair against a mix, but it does not make difficult overlap disappear. Listen to wanted transients, vocal brightness and room decay. If the bleed and source share the same content at similar level, a transparent full removal may be impossible. Our [De-bleed guide](https://izotoperx.net/izotope-rx-de-bleed/) covers settings, source diagnosis and DAW workflow. The distinction matters: in an [April 2026 discussion about a supposedly removed reference option](https://www.reddit.com/r/iZotopeAudio/comments/1sz546a/removal%5Fof%5Freference%5Ftrack%5Fin%5Frx%5F12%5Fdebleed%5Fis%5Fa/?ref=izotoperx.net), the account named iZotopeOfficial pointed readers to Classic in the editor. The current manual independently confirms that retained workflow; the thread is evidence of user confusion, not a performance benchmark. ## Breath Control gets new machine-learning detection Breath Control is also rebuilt rather than newly invented. iZotope says RX 12 uses a new machine-learning system for finer breath detection in dialogue and vocals. The job is to reduce distracting inhalations without cutting aspirated consonants, phrase attacks or the natural pacing that makes a performance human. The [current Breath Control manual](https://docs.izotope.com/rx12/en/breath-control.html?ref=izotoperx.net) supports attenuation rather than an all-or-nothing erase. Target mode lowers detected breaths toward a target level; Gain applies the chosen reduction amount. The new Real-time and Offline algorithms sit alongside Classic, the RX 11 algorithm. Natural preserves ambience and Gated removes it; that switch is unavailable in Classic. For Real-time or Offline, the manual recommends selections or clips preferably at least three seconds long. Test strong breaths and soft word beginnings together. If the first syllable gates, the detector or reduction is too aggressive. The dedicated [RX 12 Breath Control tutorial](https://izotoperx.net/izotope-rx-breath-control/) separates Target from Gain decisions and explains when clip gain is still faster. A rebuilt detector can reduce manual corrections; it does not make every breath a defect. ## Music Rebalance improves separation and adds a real-time plug-in RX 12 updates the Music Rebalance neural models and provides a real-time plug-in in Standard and Advanced. The current module labels the four components Voice, Bass, Drums and Other; Drums includes percussion, and Other collects material outside the first three groups. The new Stems View route goes beyond level changes: isolate a component, repair it with another RX module, then evaluate the recombined mix. The [Music Rebalance manual](https://docs.izotope.com/rx12/en/music-rebalance.html?ref=izotoperx.net) documents Good / Real-time, Better / Offline and Best / Offline quality choices. The AudioSuite form has no pre-render Preview, and Stem Split uses 0 dB component gains regardless of the gain-slider positions. “Improved neural net” is not a universal quality percentage. Dense guitars, reverberant vocals, distorted bass and cymbal-heavy mixes can still leak across boundaries. Compare transients, stereo image and ambience at equal loudness. For operational settings, stem export and artifact checks, use the [Music Rebalance guide](https://izotoperx.net/izotope-rx-music-rebalance/). Judge the new model on the actual recording; a clean solo stem does not guarantee a clean recombined mix. ## Dialogue Isolate receives updated neural separation Dialogue Isolate remains the speech-versus-background tool, but RX 12 updates its neural network. It separates dialogue, noise and reverb and can send those components into Stems View. That matters when a result needs more than one global Voice/Noise/Reverb balance—for example, when a short siren leaks into speech or a reflection requires local work. The current [Dialogue Isolate manual](https://docs.izotope.com/rx12/en/dialogue-isolate.html?ref=izotoperx.net) lists Voice, Noise, Reverb and Sensitivity in Standard and Advanced. The Quality menu and four-band controls are Advanced-only in both the module and plug-in. Best / Offline is intended for the RX Audio Editor and offline DAW bounces or renders; it is not an Advanced editor-only capability. Higher sensitivity can classify more source as noise or reverb, increasing artifacts. The new model still requires a representative quiet line and a difficult overlap test. Use our [Dialogue Isolate workflow](https://izotoperx.net/izotope-rx-dialogue-isolate/) for the actual settings and stop conditions. If the problem is steady room noise rather than overlapping non-stationary material, a denoise module may remain the simpler answer. ## Repair Assistant is the main RX 12 change for Elements Elements does not receive Scene Rebalance, Stems View, Trim Silence, De-bleed or Breath Control. The official feature comparison does list an improved Repair Assistant neural model across Elements, Standard and Advanced. That is the relevant RX 12 change for an Elements user: better assistive analysis inside the limited plug-in set, not access to the full editor inventory. Repair Assistant proposes a starting chain. It does not know the story priority, whether a breath is intentional, or which alternate microphone is available. Use its section On/Off controls to audition each contribution and reduce any stage that changes speech identity. The [Repair Assistant manual](https://docs.izotope.com/rx12/en/repair-assistant.html?ref=izotoperx.net) distinguishes these internal controls from Open Module Chain, which is available only in the Advanced editor. Elements users work in the plug-in; the full editor workflow does not come with that edition. For ownership decisions, the [RX Elements, Standard and Advanced comparison](https://izotoperx.net/izotope-rx-elements-vs-standard-vs-advanced/) maps the current boundaries. ## The editor is faster to navigate RX 12 enlarges the default spectrogram, adds a module search bar and makes the History panel resizable. The larger working area helps precise selections on a laptop. Search reduces menu hunting when the module list is long. A larger history view makes it easier to identify and return to a specific repair state. These changes improve operating speed, not sound by themselves. Test them during a real edit: find five modules without using the category list, compare three History states, and make a tight selection around a transient. The [RX selection and History guide](https://izotoperx.net/izotope-rx-selections-preview-compare-history/) turns those conveniences into a repeatable audition process. ## What is new in each RX 12 edition? **RX 12 Elements:** the improved Repair Assistant neural model is the main version change. Elements remains a compact plug-in package rather than the standalone spectral workflow used throughout the advanced tutorials. **RX 12 Standard:** Stems View, rebuilt ML De-bleed, rebuilt ML Breath Control, a real-time De-bleed plug-in, a real-time Music Rebalance plug-in, and improved Music Rebalance and Dialogue Isolate models form the core delta. Standard also receives the editor design updates. **RX 12 Advanced:** everything in Standard, plus the new Scene Rebalance and Trim Silence modules. Advanced is the only edition that can build dialogue/music/effects stems from a flattened scene through Scene Rebalance. It also retains the specialist repair modules that were already Advanced-only. Do not infer edition ownership from a screenshot or a bundle name. Check the current iZotope comparison immediately before purchase or deployment because account entitlements and bundles are separate from the module matrix. ## What RX 12 did not make obsolete Core diagnosis still matters. De-click remains for clicks, De-clip for overload, De-hum for tonal interference, De-plosive for microphone pops and Spectral Repair for selected events. A new stem layer does not mean every defect should pass through source separation first. RX 11 sessions and proven chains do not become wrong because RX 12 exists. The upgrade question depends on whether the new routes remove a recurring bottleneck. This article intentionally stops at the feature map; the RX 11 comparison handles migration, compatibility risk and purchase value. ## Eight steps to test RX 12's new features 1. **Confirm RX 12 and your edition.** Record the installed RX 12 version and identify whether the license is Elements, Standard or Advanced; compare it with the current release notes. 2. **Choose representative source files.** Use a flattened scene, music mix, bleed problem, breath-heavy voice file and long-form recording only when those sources match your work. 3. **Test the headline workflow first.** Choose a workflow your edition includes: Scene Rebalance or Trim Silence in Advanced, Stems View in Standard or Advanced, or Repair Assistant in Elements. Test Trim Silence on a separate whole-file copy. 4. **Test rebuilt modules separately.** In Standard or Advanced, evaluate De-bleed and Breath Control against unprocessed duplicates before putting them inside a long chain. 5. **Check neural-separation updates.** In Standard or Advanced, compare Music Rebalance and Dialogue Isolate on difficult overlap, listening for leakage, transient damage and speech texture. 6. **Evaluate the editor changes.** In the Standard or Advanced editor, use module search, the larger spectrogram and resizable history during a real repair. 7. **Compare at equal loudness.** Audition from the same pre-roll so louder playback does not masquerade as cleaner processing. 8. **Record a workflow verdict.** Note the edition, file, artifacts, time saved and follow-up corrections for each feature before changing production practice. An example of a useful evaluation note would be: “Stems View let me repair bass hum without dulling the vocal, but recombination needed one automation fix.” A vague “RX 12 sounds better” cannot guide an edition choice or a repeatable workflow. Keep original audio and an RX Document before committing edits in the editor. The [RX file guide](https://docs.izotope.com/rx12/en/working-with-files.html?ref=izotoperx.net) explains that an RX Document retains the original and edit history, while ordinary Save on an opened WAV or AIFF overwrites that file. Export separate delivery audio and reimport it to check duration, channel order and alignment. For rendered comparisons use History or an untouched copy: [module Bypass operates during Preview](https://docs.izotope.com/rx12/en/common-module-controls.html?ref=izotoperx.net) and does not undo earlier renders. ## Quality control for the new separation workflows - Keep an untouched duplicate aligned with the processed file and solo one version at a time when comparing. For time-removing operations such as Trim Silence, review pacing separately and verify the new duration. - Match loudness before comparing new and old models. - Check leakage between stems, not only the soloed target. - Listen to transients, sibilants, reverb tails and low-level ambience. - Recombine stems before approving a repair. - Check local edit boundaries and final export length. - Record the exact edition and processing mode. - Retain the RX 11 or pre-RX 12 route until the new workflow passes real deliverables. The [Learn iZotope RX hub](https://izotoperx.net/learn-izotope-rx/) connects each new feature to its working tutorial, while the review and version-comparison pages handle buying intent separately. ## Frequently asked questions ### What are the biggest new features in iZotope RX 12? The three headline additions are Scene Rebalance, Stems View and Trim Silence. RX 12 also rebuilds De-bleed and Breath Control, improves several neural models and updates the editor interface. ### When was iZotope RX 12 released? The release notes checked September 7, 2026 UTC list RX 12.0.0 as the initial major release on April 28, 2026 and show no later RX 12 entry. ### Is Scene Rebalance included in RX 12 Standard? No. Scene Rebalance is an RX 12 Advanced feature. It separates dialogue, music and effects from a single stereo render and can send those components to Stems View. ### Is Stems View included in RX 12 Standard? Yes. Stems View is available in Standard and Advanced. Standard can enter it through Music Rebalance and Dialogue Isolate; Advanced also adds the Scene Rebalance route. ### What changed in RX 12 De-bleed? Standard and Advanced gain reference-free machine-learning De-bleed and a real-time plug-in. The Audio Editor still includes Classic, the RX 11 reference-based algorithm; Classic is absent from the new plug-in. ### What changed in RX 12 Breath Control? Breath Control was rebuilt with new machine-learning detection intended to identify breaths more accurately in speech and vocals. It remains a tool to reduce breaths carefully, not a reason to erase every inhale. ### Does RX 12 Elements get any new features? Its main version improvement is the updated Repair Assistant neural model. Elements does not include Scene Rebalance, Stems View, Trim Silence, De-bleed or Breath Control. ### Do the new features make RX 12 worth upgrading from RX 11? Only when they solve recurring work and fit your edition and system. Test representative files, measure time saved and check artifacts; use the dedicated RX 12 versus RX 11 guide for the full upgrade decision.