On this page
RX 12 De-plosive reduces low-frequency microphone bursts such as P-pops. Start with enough Strength to hear the effect in Preview, tune Sensitivity to catch the real events, then balance Strength and Frequency Limit against the clean voice. A quieter pop is useful only if the word still sounds natural.
Use the unfiltered recording. De-plosive relies on very low frequencies for detection, so a high-pass filter applied first can hide the evidence it needs. Run De-plosive before broad EQ, then spot-fix any exceptional event instead of thinning the entire voice.
De-plosive is available as an Editor module and plug-in in RX 12 Standard and Advanced; Elements does not include it. The current edition comparison confirms that boundary. The steps below use the standalone Editor; a DAW insert processes through its host rather than the Editor’s Preview and Render buttons.
What a recorded plosive sounds and looks like
iZotope's plosive-removal guide describes the problem as a blast of air striking the microphone diaphragm and creating a low-frequency bump, sometimes with distortion. P and B are familiar triggers; T, K and other forceful releases can also produce pressure changes.
On a waveform, the event can look like a large slow excursion around the consonant. In a spectrogram with enough low-frequency detail, it appears as a brief concentration near the bottom. Listen as well as look: a kick, mic bump, wind gust and digital pop can occupy a similar moment but need a different repair.
A microphone P-pop follows the air release of a consonant. Mouth clicks are usually shorter impulses; sustained rumble continues beyond the word. These are listening clues, not a diagnosis from shape alone. De-plosive can also help some mic bumps, but a wind gust or handling noise is not automatically a speech plosive. Compare the suspect event with the untreated recording before choosing a detector.
De-plosive controls at a glance
| Control | What it changes | Failure when pushed too far |
|---|---|---|
| Sensitivity | Which events are detected as plosives | Clean low speech is treated as a problem |
| Strength | Depth of attenuation on detected events | The word loses body or sounds hollow |
| Frequency Limit | Highest frequency De-plosive will reduce | Wanted vocal harmonics enter the repair |
The RX 12 De-plosive manual says Sensitivity affects overall plosive reduction more than simply increasing Strength. Sensitivity controls what is detected; Strength controls how deeply detected content is reduced. Audition them together: a missed event calls for a detection adjustment, while damaged clean speech calls for less intrusive processing.
Nine steps to remove P-pops safely
- Preserve the source. Keep an independent unfiltered audio backup and open a separate working file. A duplicate clip or playlist may still reference the same media.
- Confirm the artifact. Listen for a brief microphone air burst tied to speech; distinguish clicks, sustained rumble and wind.
- Choose a representative phrase. With Instant Process off, use the Time selection tool [T] to include the pop, clean low voice, vowels and nearby consonants.
- Set detection first. Use Preview with modest audible Strength and raise Sensitivity only as far as the target needs.
- Set the boundary. Place Frequency Limit just above the burst, then check overlapping voice body and harmonics by ear.
- Balance attenuation. Adjust Strength for a less distracting boom. Compare a higher-Sensitivity/lower-Strength candidate if that preserves speech better.
- Compare candidates. Use Preview/Bypass or Compare on the same untreated passage at matched listening level; do not stack test renders.
- Spot-repair leftovers. Render the chosen repair on the intended range, then assess any remaining event locally. Stop if another pass damages the word.
- Save and approve in context. Export to a new audio filename, reopen it and check the real dialogue or vocal chain for body, timing and intelligibility.
No number is universal. Voice pitch, microphone, distance, angle, wind protection and the intensity of the air blast change where the event ends. Use published settings as starting examples, not facts about your recording.
The Editor selection tools distinguish making a selection from Instant Process, which applies a configured repair as soon as you select. Keep Instant Process off here. Confirm the intended file tab and time range before Render; do not leave several files in Composite View for a one-voice repair.
Build a test loop before processing
Make a short audition loop with three parts: the obvious P-pop, the cleanest low vowel from the same speaker, and the consonant transition immediately after the damaged event. This is more revealing than looping the boom alone. A setting can make the defect disappear while quietly shaving weight from every nearby syllable.
Use the Preview, Bypass and Compare controls to hear restrained candidates without committing each experiment. Start each candidate from the same untreated selection. First change one control at a time; then compare the documented higher-Sensitivity/lower-Strength combination with your initial balance. Render only the candidate you choose. Compare and Preview alone do not apply the final edit.
Use both full-range headphones and speakers with credible low-frequency extension. Small laptop speakers may hide the original burst and the damage caused by repair. The goal is not a perfectly empty low-frequency spectrogram: it is a believable word that no longer makes the listener notice the microphone.
Sensitivity finds the event
Raise Sensitivity while looping a phrase that contains both a clear plosive and clean low-frequency speech. Stop when the target is found consistently. If nearby vowels or low consonants change even before much Strength is applied, lower Sensitivity or shorten the selection.
iZotope documents a 20–80 Hz detection range and warns that prior high-pass filtering may prevent correct detection. If the filter is still adjustable, bypass it before repair. If filtering was baked into a rendered file or the recording hardware, return to an unfiltered backup when available. Boosting bass afterward does not recreate discarded information; with no original, try a restrained local EQ or spectral repair and accept that some damage may remain.
For a long voiceover, a conservative global detection pass can catch ordinary P-pops. Leave unusual events for local selections; do not raise Sensitivity across an hour just to catch one consonant.
The current RX 12 Standard feature page describes targeted dialogue/vocal repair and mentions mic bumps as well as plosive pops. Use that as a reason to test a suitable event, not to put a strong detector across music, ambience or a finished mix by default.
Strength determines how deeply it is reduced
Strength acts after detection. Increase it until the low-end burst no longer pulls attention from the word. Then compare with the untreated phrase at matched loudness. The repair should sound like the same speaker at the same distance—not a different, thinner voice for one syllable.
The manual warns that high Strength can reduce speech quality and suggests trying higher Sensitivity with lower Strength to preserve it. That is an alternative to audition, not permission to maximize detection: excessive Sensitivity can also catch clean speech. Compare chest resonance, vowel weight and the transition into the next sound, and retain whichever restrained balance leaves the word more believable.
A large residual burst can trigger downstream compression or limiting and cause an audible level fluctuation. Check the repaired phrase through the actual final chain, then bypass stages to find what changes it. Repairing the burst before dynamics is a practical starting order, not proof that every exported pumping artifact comes from compression.
Frequency Limit protects the voice
Frequency Limit sets the upper boundary for plosive reduction. Use the burst’s visible extent as a starting point and set the boundary just above it. Plosive energy can overlap the voice’s fundamental and harmonics, so there may be no clean frequency line that separates damage from speech. Check clean vowels and use a local time selection when a wider boundary harms them.
The RX 12 manual describes plosives typically between 20 and 300 Hz, sometimes reaching 500 Hz. These describe possible signal content, not mandatory settings or the detector band. Several hundred hertz can include useful vocal weight, so raise the boundary only when that recording justifies it.
In iZotope’s voice-recording tutorial, 120 Hz is one narrator’s example. It is not a prescription for every voice or microphone. That older tutorial uses an earlier interface; follow the current Editor’s Render control to apply an approved repair.
De-plosive versus a high-pass filter
A high-pass filter attenuates low frequencies whether the plosive is present or not. It can quickly reduce one boom, but a fixed track-wide filter also reduces legitimate vocal body. De-plosive is designed to detect events and act when they occur.
Use De-plosive before a broad high-pass when you still have unfiltered material. Then decide whether unrelated rumble needs global EQ. For a stubborn isolated P-pop, test a short automated high-pass or local clip EQ; listen at both boundaries and use fades where needed. Moving an EQ cutoff and changing clip gain are different operations: gain lowers the whole selected signal, not only its bass.
The RX De-hum guide covers persistent tonal low-frequency interference. A plosive is a brief pressure event, not a mains fundamental and harmonic ladder.
Whole-file versus spot processing
Whole-file De-plosive is efficient when one speaker, microphone and distance remain stable. Tune on several phrases: the strongest pop, a low clean sentence and a quiet line. Keep the global pass gentle enough that none of them loses body.
Spot processing is safer when only a few events are severe or the speaker moves around the microphone. Select each event with a small amount of surrounding speech, tune a higher Frequency Limit or Strength locally, and compare the transition.
iZotope's current recorded-vocal guide recommends conservative settings and local repair for severe problems. That is the practical compromise: automation for ordinary events, attention for exceptions.
When De-plosive makes the voice thin
First lower Strength. If the word is still hollow, lower Frequency Limit so fewer harmonics are affected. If processing triggers on clean low speech, reduce Sensitivity or restrict the time selection. Recompare at matched loudness after each change.
Check whether an EQ or denoiser later in the chain is responsible. De-plosive plus a steep global high-pass plus strong noise reduction can remove the same low-frequency body three times. Bypass stages one by one using the RX Module Chain workflow.
If the remaining event is mainly midrange distortion rather than low-frequency air, stop increasing De-plosive. Select the artifact in the Editor and test Spectral Repair.
De-plosive versus De-click and Mouth De-click
Use De-plosive for the low-frequency air burst. Use general De-click for digital impulses, pops and discontinuities. Use Mouth De-click for saliva clicks and lip smacks. Similar words in the editor's notes do not make the events equivalent.
A complex consonant can contain a low P-pop and a mouth click. Compare the dominant problem first, then treat the second only if it remains audible; there is no universal requirement to run both in one fixed order. The Mouth De-click tutorial explains that module’s removed-signal monitoring. Do not assume every RX module has the same monitoring controls.
Do not run three aggressive global detectors by default. Confirm each problem, audition the result and leave a module bypassed when it does not solve an audible defect.
Voiceover, podcast and vocal workflows
For voiceover and podcasts, a useful starting point is plosive repair before broad denoise, EQ and compression. iZotope’s plug-in versus Editor guidance recommends that order for De-plosive, while also advising local adjustments for exceptions. Plan your take edits as the production requires; processing order should respond to the defects in the source.
If the recording also carries steady room noise, audition the plosive repair on the untouched file, then recheck after denoising. iZotope’s podcast and audiobook cleanup workflow separates restoration, editing and output decisions; it is not a requirement to run every processor. A chain preset cannot decide whether a particular word needs repair.
For sung vocals, preserve intentional consonant impact and musical low-mid weight. Spot-process severe bursts when the singer changes distance or intensity. A global setting approved on a quiet verse may overreact in the chorus.
The RX vocal-cleanup guide places plosive repair among mouth noise, breath, sibilance, background and room decisions without making every processor mandatory.
Prevent plosives before recording
Shure’s microphone placement and pop-filter guidance recommends moving the capsule out of the direct air stream and putting a pop filter between mouth and mic. Try a slightly off-axis position and suitable distance, then monitor repeated P and B sounds. No one distance or filter guarantees a clean take; check both pops and the resulting voice tone.
For a narrator, mark recurring problem words and adjust angle rather than backing so far away that room sound replaces plosives. For a singer, preserve technique and emotional delivery; find a position that breaks the air blast without making the performance timid.
If a retake is available and the repair changes intelligibility or timbre, record the line again. Restoration is most valuable when the take cannot be recreated.
Troubleshooting De-plosive results
| What you hear | Possible cause to test | Next move |
|---|---|---|
| The P-pop is unchanged | Sensitivity is too low, or a prior high-pass filter removed detector evidence | Return to the unfiltered source and tune detection first |
| The pop is quieter but still distorted | The event may include distortion beyond the low-frequency burst | Keep the restrained De-plosive pass and spot-test Spectral Repair |
| Every low word sounds smaller | Sensitivity, Strength or Frequency Limit may be excessive | Lower the boundary or process only marked events |
| The repaired syllable pumps after export | Dynamics processing or the repair boundary may expose a discontinuity | Recheck through the final chain and soften the repair |
| A click remains at the consonant | An original click or a new processing artifact may remain | Confirm it separately before using Mouth De-click or De-click |
When the cause is uncertain, duplicate the event and test the competing diagnosis on separate copies. A wrong processor can make a symptom smaller while creating a new artifact; an A/B comparison exposes that tradeoff quickly.
Render and save a repaired copy
In the standalone Editor, Render changes the open audio; saving to disk is a separate action. The RX saving and export reference says Save overwrites an uncompressed WAV/AIFF original. Export to a new filename, or use Save As for an uncompressed working copy, and keep the untouched backup separate.
Save an RX Document (.rxdoc) if you need the original and edit history for later work inside RX. Deliver a supported audio format such as WAV rather than the RX document. Reopen the exported file and compare the repaired words, clean speech, duration and channel layout. Do not shorten the timeline to remove a pop when sync with video or other tracks must remain intact.
Final De-plosive quality gate
- The event is a speech-related low-frequency air blast, not a click, hum or sustained wind.
- De-plosive runs on unfiltered audio before a broad high-pass filter.
- Sensitivity was set for detection before Strength was increased.
- Frequency Limit follows the actual burst instead of a universal preset.
- Clean low voice, vowels and consonant transitions remain natural.
- Exceptional events are spot-processed rather than forcing stronger global settings.
- The full chain was checked for stacked low-frequency loss.
- The repaired phrase was approved in the real dialogue or vocal mix.
The Clean Dialogue with RX hub connects plosive repair to mouth clicks, noise, reverb and other speech-specific problems.
Frequently asked questions
What does RX De-plosive remove?
It detects and attenuates brief low-frequency pressure bursts created when speech air hits the microphone, commonly heard as P-pops.
What is the best De-plosive Frequency Limit?
There is no universal value. Start just above the actual burst, then check the voice by ear because plosive energy and vocal harmonics can overlap. Use a local time selection if a broad setting thins clean speech.
Should I use De-plosive before a high-pass filter?
Yes. iZotope says prior high-pass filtering can impair the 20–80 Hz detection band. Repair first, then apply any justified global EQ.
Why does De-plosive make my voice thin?
Strength, Frequency Limit or Sensitivity may be too high, or a later high-pass/denoiser is stacking low-end reduction. Lower one control at a time and bypass the chain.
Can a high-pass filter remove P-pops?
It can reduce low-end boom, especially as a local automated fix, but a fixed track-wide filter also removes wanted low-frequency voice body.
Should I process the whole voice track?
Use a conservative global pass only after representative clean and damaged phrases pass listening checks. Stable speaker and microphone conditions help; severe or unusual events still need local attention.
Is De-plosive included in RX 12 Elements?
No. iZotope's current comparison lists De-plosive in RX 12 Standard and Advanced, not Elements.
When should I use Spectral Repair instead?
Use it when the leftover is irregular, extends into the midrange or remains only in one local event that De-plosive cannot reduce without thinning speech.
