> ## Content Index > Fetch the complete content index at: https://soundforgepro.com/llms.txt > Use this file to discover other available public pages before exploring further. # iZotope RX De-hum in RX 12: Static vs Dynamic, Settings and Test Results - URL: https://soundforgepro.com/izotope-rx-de-hum/ - Published: 2026-10-25T07:41:00.000Z - Updated: 2026-10-25T07:40:59.000Z - Description: RX 12 De-hum has two modes and an Adaptive switch. Which one fits hum, buzz or drifting tones, a Dynamic routine, the Static settings that keep a voice intact, edition and host limits, and what published reviews found. - Author: Erick Finn - Tags: iZotope RX, Sound Forge Restoration Guides **Use De-hum's Static mode for a steady hum with only a few harmonics that sits below the voice, Dynamic mode with Learn for buzz, long harmonic ladders or hum that overlaps speech, and Adaptive mode, rendered offline, when the hum drifts in pitch through the file. In RX 12, De-hum is an RX Audio Editor module in Standard and Advanced, a plug-in in all three editions including Elements, and a processing mode inside the Spectral Editor ARA plug-in in Standard and Advanced.** The [RX 12 De-hum manual page](https://docs.izotope.com/rx12/en/de-hum.html) describes the module as a remover of persistent tonal noise: AC hum, electrical buzz and interference from electromagnetic fields. My own order of attack is short. I'd start in Dynamic, learn from a few seconds of hum with no voice on it, and raise Sensitivity only until the hum stops showing up under the words. I'd switch to Static when the hum is a clean, stable tone well below the dialogue, because a handful of fixed notches is easy to audit by ear. Adaptive is my fallback for files where Dynamic works in one place and fails in another, not a starting point. Everything below refers to RX 12; the [RX release notes](https://www.izotope.com/pages/release-notes/rx-advanced) date version 12.1.0 to October 15, 2026 and list no De-hum changes in it. If RX itself is new to you, the [RX 12 tutorial for beginners](https://soundforgepro.com/izotope-rx-12-tutorial-beginners/) covers the editor basics first. ## What De-hum removes, and what it leaves behind Hum is tonal. It has a fundamental, usually tied to the mains frequency, and a ladder of harmonics stacked above it. iZotope's [technical guide to RX De-hum](https://www.izotope.com/community/blog/deep-dive-into-rx-de-hum) puts typical fundamentals at 50 or 60 Hz and notes that 16 harmonics on those fundamentals cover the range up to roughly 1 kHz, while buzz can run past 1 kHz, even to 10 kHz, with hundreds of harmonics. That split between a short ladder and a very long one is the reason the module has two modes. The causes are mostly electrical. An [iZotope walkthrough on removing audio hum](https://www.izotope.com/community/blog/removing-audio-hum-with-izotope-rx) lists poor grounding of computer peripherals, outlets and power supplies, plus fans, air conditioners, electric heaters and light switches. A later [iZotope article on high-frequency buzz](https://www.izotope.com/community/blog/removing-high-frequency-buzz-and-interference-from-audio) adds cabling faults: grounding problems, electromagnetic interference, poor shielding, bad connections and long runs of audio cable wound together with power cable, especially on unbalanced lines. If a retake is still possible, fixing the cable or the ground beats any repair. What De-hum leaves alone is the random part of the noise. Hiss, room tone and traffic are broadband, so they are still there after the hum is gone; in iZotope's 2021 walkthrough the leftover background noise went to Dialogue Isolate or Voice De-noise in a second step. A single stray tone that comes and goes is also a poor fit for the module. Both cases are covered in the module table near the end of this guide, and the wider order of operations sits in the [background-noise cleanup across RX modules](https://soundforgepro.com/remove-background-noise-izotope-rx/). This page stays with hum and buzz, one part of the [iZotope RX toolkit](https://soundforgepro.com/izotope-rx/). ## Static, Dynamic or Adaptive: pick the mode first The mode decides more than any slider does. The manual's guidance is brief: Static for simple tonal noise with only a few harmonics of the fundamental and for hum that does not overlap speech; Dynamic for hum with more than a dozen harmonics, inharmonic hum made of unrelated tones, buzz or EMF that reaches the upper frequencies, and hum or buzz that overlaps speech. Adaptive is a switch on top of that: it lets the notches follow changes in the audio over time and works better on hum that changes pitch. The [RX Standard product page](https://www.izotope.com/products/rx-standard) quotes up to 1,024 dynamically variable notch filters for Dynamic mode. | Mode | Reach for it when | How it filters | What to watch | | --------------------- | ------------------------------------------------------------------------- | -------------------------------------------------------------------- | --------------------------------------------------------------------------------- | | Static | A steady hum with a few harmonics, sitting below the voice | Fixed notches on the fundamental and up to 16 harmonics | Notches that land inside the voice thin it out and can ring at high Q | | Dynamic | Buzz, long or uneven harmonic ladders, hum under speech | Up to 1,024 gated notches driven by a learned hum profile | Needs a clean Learn region; more Bands costs transparency | | Dynamic with Adaptive | Hum that drifts in pitch or level, such as edits built from several takes | Notches follow a hum profile that keeps updating, with no Learn step | High latency and offline-only in the plug-in; can mistake sustained notes for hum | Why Dynamic sounds cleaner on dialogue comes down to what iZotope calls gated notching. The algorithm watches the energy at each frequency: when the signal looks like hum, the notch cuts, and when the energy rises above what the hum should produce, the notch stops attenuating. Fixed notches cannot do that. In the speech example from iZotope's technical guide, Static carved holes into the speech harmonics between 300 and 700 Hz, while Dynamic left them intact and also pulled down buzz around 8 kHz. The same article explains when Static is still safe. If the speech sits entirely above 150 Hz, notching hum at 60 and 120 Hz will not make it ring; notch more than two harmonics and you are cutting into the voice. For the Dynamic-or-Adaptive call, iZotope's rule is practical. Use plain Dynamic when the hum is stable in frequency and level, especially when you have a few seconds of pure hum to learn from. Move to Adaptive when Dynamic clearly treated some parts of the file better than others, which suggests the hum is wandering. Files spliced from several takes are a common case, because Dynamic's narrow notches react to even small shifts in the hum's frequency. Keep Adaptive away from music where you can: iZotope built it mainly for speech and warns that sustained notes can be taken for hum. ![Decision chart starting from the question What kind of hum or buzz with five rows: steady hum with few harmonics below the voice leads to Static, many or unrelated harmonics, high buzz or hum under speech leads to Dynamic plus Learn, hum that drifts or returns leads to Dynamic plus Adaptive rendered offline, one stray tone now and then leads to Spectral Repair Attenuate Vertically, and hiss left after De-hum leads to Spectral De-noise or Voice De-noise](https://storage.ghost.io/c/af/4b/af4b05d4-2e0b-4edc-8378-1e2a62fecc70/content/images/2026/10/de-hum-mode-decision.webp) Pick the mode from the hum itself; the last two rows are jobs for other RX modules. ## Where De-hum runs in RX 12 De-hum is one of the few RX modules every edition gets, but not every edition gets it in every place. The [RX 12 overview and edition tables](https://docs.izotope.com/rx12/en/rx-overview.html) describe Elements as six plug-ins without the RX Audio Editor, list the De-hum plug-in in Elements, Standard and Advanced, and list the De-hum module in the Audio Editor for Standard and Advanced. The Audio Editor's Spectrum Analyzer, which the manual suggests for finding the base frequency, follows the same Standard and Advanced split. | Where you use it | RX 12 Elements | RX 12 Standard | RX 12 Advanced | | ---------------------------------------------- | -------------------- | -------------- | -------------- | | De-hum plug-in in a DAW or NLE | Yes | Yes | Yes | | De-hum module in the RX Audio Editor | No (no Audio Editor) | Yes | Yes | | De-hum mode in the Spectral Editor ARA plug-in | No | Yes | Yes | | Spectrum Analyzer for finding the fundamental | No | Yes | Yes | RX 12 plug-ins come in AU, AAX, AAX AudioSuite and VST3, with AU ARA and VST3 ARA used by the Spectral Editor, all 64-bit only, according to the [RX Advanced product page](https://www.izotope.com/products/rx-advanced). The [RX Elements product page](https://www.izotope.com/products/rx-elements) gives the confirmed RX 12 host list: Logic Pro 12, Pro Tools 2026, Ableton Live 12, Cubase 15, Nuendo 15, Fender Studio Pro 8, Reaper 7, FL Studio 26, Adobe Audition 2026, Premiere Pro 2026, DaVinci Resolve 19 and Reason 13\. Sound Forge Pro is not on that list, so for hum work inside Sound Forge itself the [Sound Forge hiss and hum guide](https://soundforgepro.com/remove-hiss-hum-sound-forge/) is the better route. Two settings decide whether De-hum works as a live insert. Adaptive Dynamic mode is not optimized for real-time use, and the manual tells plug-in users to run it offline; the RX Standard product page adds that Adaptive has high latency and works best in the standalone app and offline plug-in formats. When Adaptive arrived in RX 10, iZotope's engineer put the plug-in's latency in that mode at roughly 2,000 ms because of the lookahead it needs. In Static, Linear-phase filters add latency too, and turning them off reduces it when De-hum runs as a real-time plug-in. The third surface is easy to miss. In the [RX Spectral Editor ARA plug-in](https://docs.izotope.com/rx12/en/rx-spectral-editor.html), De-hum is a processing mode next to Attenuate, Replace, Gain and De-click. It runs Dynamic mode with Adaptive learning switched on and exposes only Sensitivity, Bands and Filter Q, so there is no Static ladder to tune. iZotope's edition tables list ARA support for Logic Pro 10.7 or later (AU, in Rosetta only) and for Studio One 7 and Studio Pro 8 (VST3). The edition question goes well beyond De-hum, and the [RX Elements vs Standard vs Advanced comparison](https://soundforgepro.com/izotope-rx-elements-vs-standard-vs-advanced/) covers it module by module. For context, De-hum has been part of RX since version 1; Dynamic mode arrived in RX 9, when the original method became Static, and Adaptive followed in RX 10. ## Dynamic De-hum, step by step This is the routine I would use on a dialogue clip with hum or buzz under it. It is the same in the Audio Editor module and in the plug-in, apart from how Learn is triggered. 1. **Find hum on its own.** Look for a pause, room tone before the first line or a tail after the last one. Learning from a region with voice in it is the usual reason Dynamic underperforms; the manual says RX can analyze any audio with prominent hum, but the results may not be as useful. 2. **Set Filter type to Dynamic and learn.** In the Audio Editor, select the hum and click Learn. In the plug-in, enable Learn and play the hum section back. 3. **Bracket the hum with the range selectors.** The two handles on the display set the lower and upper bounds of processing. A low hum with no buzz does not need the range open to the top of the spectrum. 4. **Raise Sensitivity gradually.** Sensitivity sets how much hum is removed. Bring it up while previewing a section with speech, and stop when the hum disappears under the words, not when the pauses go dead silent. 5. **Add Bands only for complex hum.** Bands sets the number of dynamic notch filters. Higher values handle dense buzz, and the manual is explicit that they cost transparency. 6. **Touch Filter Q last.** It sets the notch bandwidth. Narrow notches leave more of the voice alone; widen them only if hum still leaks through. 7. **Check, then render.** Toggle Bypass during Preview on a passage where voice and hum overlap. If the hum survives in some sections and not in others, switch on Adaptive mode and render offline. After rendering, hiss and room noise will still be there, and that is expected. iZotope positions Dynamic De-hum as the step before Voice De-noise or Spectral De-noise, which deal with the random part of the noise. ![RX 12 De-hum panel in Dynamic mode with a Learn button and an Adaptive mode checkbox above a spectrum where five narrow notches cut low hum peaks between two range handles, Sensitivity, Bands and Filter Q sliders below it and footer buttons Preview, Bypass, Compare and Render; orange markers 1 to 4 point to Learn, the Dynamic filter type, Sensitivity and Render](https://storage.ghost.io/c/af/4b/af4b05d4-2e0b-4edc-8378-1e2a62fecc70/content/images/2026/10/de-hum-dynamic-controls.webp) Learn on pure hum (1) with Filter type set to Dynamic (2), raise Sensitivity (3) only until the hum drops under the voice, then Render (4) after a Bypass check. ## Static mode: the settings that matter Static is the older, more manual method, and it rewards patience. The RX 12 manual lists the controls below for the Static view, where the panel labels the bandwidth slider simply "Q". The Static view also carries Learn and an Adaptive mode checkbox. Here Learn sets the base frequency from the learned profile, which is usually the fastest way to land on the right fundamental. | Control | What it does | How I would set it | | ----------------------------- | --------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------- | | Frequency | Sets the fundamental of the hum | Let Learn set it, then nudge it while previewing; trust the detected value over a 50 or 60 Hz assumption | | Q | Bandwidth of the notches on the fundamental and every harmonic | Narrow enough to miss the voice; a higher Q narrows the notch but adds ringing | | Harmonics | Adds notches above the fundamental, up to 16 | Start with few and add until the visible ladder is covered; stop before the notches reach dense speech | | Link harmonics | All (one gain node, the default), Odd/Even (two nodes) or None (one node per notch) | All for an even ladder, Odd/Even when odd and even harmonics differ in level, None for one or two hot peaks | | Slope | With linked harmonics, sets how aggressively the higher harmonics are cut; their gain moves toward 0 dB as the order rises | Use it to focus the cut on the root tone and go easier on the upper harmonics, where the voice lives | | Harmonic panel (Enable, Gain) | Switches single notches on or off and sets each notch level | Disable notches where there is no hum; iZotope notes this helps preserve the original sound | | Linear-phase filters | FIR filters with no phase change, at the cost of latency and pre-ringing; off means minimum-phase IIR filters that only post-ring | Leave off for dialogue; iZotope says minimum phase is usually preferred when ringing is the problem | | High-pass and low-pass | Conventional filters ahead of the notches, useful for extreme hum or buzz | At a high Q in the default IIR mode they can resonate at the cutoff; Linear-phase mitigates that | | Filter DC offset | Removes DC offset from converters or analog circuits | Tick it when the waveform sits off the center line | iZotope's 2023 buzz article shows the Static path on real material. A shotgun microphone with a grounding fault produced hum around 120 Hz plus harmonics; because the harmonics sat at fixed intervals and did not change pitch, the editor used Learn and Static mode, then cleared the remaining high buzz with Spectral De-noise. Note the frequency: 120 Hz, not the 50 or 60 Hz a regional rule of thumb would suggest. Let Learn or an analyzer find the fundamental rather than assuming it. If the offset tick box is new to you, this walkthrough on [spotting and removing DC offset](https://soundforgepro.com/remove-dc-offset-sound-forge/) shows what the problem looks like on a waveform. Elements users have no Audio Editor and so no built-in Spectrum Analyzer. The plug-in's Learn still sets the frequency, and a host analyzer or a browser-based [online spectrum analyzer with FFT peak readout](https://soundforgepro.com/audio-spectrum-analyzer/) will show where the hum peaks sit before you start adding harmonics. ![RX 12 De-hum panel with Filter type set to Static, Linear-phase filters and Filter DC offset checkboxes, a spectrum display with a row of harmonic notches from about 120 Hz that get shallower and crowd closer together toward 1 kHz, Frequency and Q sliders, a Harmonics slider, a Link harmonics menu set to All, a Slope slider and footer buttons Preview, Bypass, Compare and Render; orange markers 1 to 4 point to Frequency, Harmonics, Link harmonics and Slope](https://storage.ghost.io/c/af/4b/af4b05d4-2e0b-4edc-8378-1e2a62fecc70/content/images/2026/10/de-hum-static-harmonics.webp) Set the fundamental (1), add only the harmonics you can see (2), choose how their gains link (3) and use Slope (4) to go easier on the upper notches. ## Hear what you are removing The most useful check in Static is Output Hum Only. With it ticked, Preview plays only the material De-hum is taking out. iZotope's routine is to find a section where the hum is mixed with other material, enable the option, click Preview and adjust Filter Q and Slope until you hear as much hum and as little program as possible. If words are clearly audible in that hum-only signal, the notches are cutting the voice: narrow them, ease off the upper harmonics with Slope, or disable the notches that land in speech. Dynamic works differently. The RX 12 manual documents Output Hum Only only among the Static controls, so I would not count on a hum-only view in Dynamic. There, toggle Bypass during Preview and listen to the places where voice and hum overlap. For high buzz, pay attention to s and t sounds: iZotope's 2023 article points out that high-frequency buzz lives in the same part of the spectrum as that diction, so heavy processing shows up there first. ## What to run after De-hum, or instead of it De-hum rarely finishes a repair by itself. These are the RX 12 modules iZotope points to around it, with the editions that include them according to the RX 12 edition tables. | What is left | Module and setting | Editions | | -------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------- | --------------------------------- | | Hiss or steady noise under speech | Voice De-noise | Elements, Standard, Advanced | | Tonal residue plus hiss, or no time for two passes | Spectral De-noise, with separate tonal and random reduction | Standard, Advanced | | Traffic, crowd or room sound around dialogue | Dialogue Isolate | Standard, Advanced | | One stray tone now and then | [Spectral Repair](https://soundforgepro.com/izotope-rx-spectral-repair/), Attenuate Vertically, Bands between 1024 and 4096 | Standard, Advanced (Audio Editor) | | Amp buzz on a guitar track | Guitar De-noise | Standard, Advanced | | Very harmonically rich buzz | De-click in Single-band or Multiband (periodic clicks) mode | Elements, Standard, Advanced | Spectral De-noise deserves a note because it can stand in for De-hum. The [Spectral De-noise manual](https://docs.izotope.com/rx12/en/spectral-de-noise.html) describes separate reduction for tonal noise such as hum, buzz or interference and for random noise such as hiss, and the De-hum page says that when there is no time for a De-hum pass followed by a De-noise pass, Spectral De-noise alone can take care of the hum rather efficiently. In the 2023 buzz example, iZotope kept Reduction low and ran a couple of gentle passes rather than one heavy one, which is the habit I would copy. For Dialogue Isolate's own faders and settings, see the [RX 12 Dialogue Isolate guide](https://soundforgepro.com/izotope-rx-dialogue-isolate/). Outside RX, Production Expert's [2026 roundup of RX alternatives](https://www.production-expert.com/production-expert-1/best-izotope-rx-alternatives-for-audio-restoration-and-dialogue-cleanup-in-2026) lists dedicated hum tools such as Acoustica DeHum, SpectraLayers Pro Hum Reduction, CEDAR Debuzz and the Hum Remover in Absentia DX, and the [comparison of RX alternatives](https://soundforgepro.com/best-izotope-rx-alternatives/) weighs those suites as a whole. ## Test results: what published reviews found No reviewer has published decibel figures for De-hum's attenuation, so the composite result here is a count of verdicts. Three independent reviews assessed the reworked module when Dynamic mode arrived in RX 9, and one 2026 practitioner review covers RX 12 use. | Source | Date | Version | Verdict on De-hum | | ------------------------ | ---------- | ------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Sound On Sound | 2021-11-18 | RX 9 Advanced | Dynamic "extremely impressive" on nasty hummy buzzes and quick to set up; ringing of static high-Q notches virtually eliminated; Static could not reach buzz above 1 kHz or non-harmonic hum | | MusicTech | 2021-10-14 | RX 9 | Static's fixed gains could cut into dialogue; Dynamic gave a much more transparent reduction with less effect on speech | | MusicRadar | 2022-01-24 | RX 9 | Dynamic mode "really extends the module's capabilities", listed as a reason to buy | | Creative Field Recording | 2026-05-06 | RX 12 | De-hum used daily in sound-effects editing; available in all RX 12 editions | **Composite result:** 3 of 3 independent reviews rate Dynamic mode a major step beyond Static for complex buzz, and 2 of 3 specifically report less damage to speech or less ringing. The [Sound On Sound review of RX 9 Advanced](https://www.soundonsound.com/reviews/izotope-rx-9-advanced) called it extremely impressive on hummy buzz and said the ringing of static high-Q notches was virtually eliminated; the [MusicTech RX 9 review](https://musictech.com/reviews/plug-ins/izotope-rx-9-review/) found it much more transparent, with less effect on speech; the [MusicRadar RX 9 review](https://www.musicradar.com/reviews/izotope-rx-9) lists it among the reasons to buy. Sound On Sound also documents Static's ceiling: it could not reach buzzy components above 1 kHz or non-harmonic hum. That lines up with iZotope's own figures, 16 harmonics reaching about 1 kHz on a 50 or 60 Hz fundamental and the 300 to 700 Hz speech holes in its comparison, which I count as vendor evidence rather than independent results. On the current version, [Creative Field Recording's RX 12 review](https://www.creativefieldrecording.com/2026/05/06/rx-12-for-sound-effects/) reports using de-hum daily. Two gaps are worth naming. None of the independent reviews assessed Adaptive mode, so the guidance on it rests on iZotope's documentation. And Production Expert's 2024 [dialogue-restoration listening test](https://www.production-expert.com/production-expert-1/results-dialogue-restoration-listening-test-noise-reduction) included a hum clip, but it compared machine-learning dialogue tools, not De-hum: Acon Extract: Dialogue, RX Dialogue Isolate and Accentize dxRevive Pro finished "practically a draw". That result says a dialogue separator can handle hum under speech; it says nothing about how De-hum compares. **My verdict:** Dynamic with a clean Learn is the right default for hum and buzz on dialogue. The reviews date from RX 9, but the RX 12 manual still describes the same two modes and the gated notching behind Dynamic, with Adaptive added on top in RX 10\. Static remains the better tool for a simple, stable hum below the voice, where a few notches are easy to check with Output Hum Only. Treat Adaptive as an offline fix for drifting hum on speech, and keep it off music with sustained notes. ## Troubleshooting thin, ringing or patchy results | Symptom | Likely cause | Fix | | ---------------------------------------- | ------------------------------------------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------------------ | | Voice sounds thin or hollow after Static | Harmonic notches land in speech; above a 60 Hz hum, more than two harmonics already cross speech frequencies | Disable notches in the harmonic panel, ease off the upper harmonics with Slope, or switch to Dynamic | | Whistling tails on consonants | Narrow, high-Q notches ring; Linear-phase adds pre-ringing | Turn Linear-phase off, lower Q slightly, or move to Dynamic's gated notches | | Hum comes back in parts of the file | The hum drifts, often across edited takes, or Learn heard voice | Relearn on pure hum; if it still varies, enable Adaptive and render offline | | Buzz above 1 kHz untouched | Static's 16 harmonics stop near 1 kHz on a 50 or 60 Hz hum | Use Dynamic with the range opened upward, or Spectral De-noise | | Sustained notes dip in music | Adaptive reads steady tones as hum | Turn Adaptive off and learn in Dynamic, or use Static on a known ladder | | Delay or sync drift in the DAW | Adaptive's lookahead, or Linear-phase latency | Render Adaptive offline; leave Linear-phase off for real-time use | | De-hum missing from the plug-in list | Format, install or scan problem in the host | Work through the fixes for [RX plug-ins not showing in a DAW](https://soundforgepro.com/izotope-rx-plugins-not-showing/) | One last habit: keep a copy of the untouched file. De-hum is easy to overdo, and a thin voice is much harder to repair than a little residual hum that the next module in the chain can still clean up.