When editing audio or video recordings, especially dialogue-heavy content, mouth noises such as clicks, pops, or lip smacks can be distracting to viewers. Removing these sounds without disrupting the natural flow of speech is a challenge faced by many editors. In this article, we explore effective strategies to eliminate mouth noises while maintaining the integrity of the dialogue.

In professional audio post-production, the presence of mouth clicks, lip smacks, tongue pops, and other oral sounds can immediately break immersion for a listener. These artifacts, while often subtle, accumulate across a long interview or narrative project and can tire the audience. The goal is not to sterilize the audio but to reduce or remove only the problematic noises so that the spoken word remains clear, natural, and engaging. Modern digital audio workstations (DAWs) and specialized tools give editors an arsenal of methods to accomplish this, but the key is knowing which technique to apply in which situation and, more importantly, how to apply it without leaving audible gaps or unnatural transitions.

Understanding Mouth Noises: Causes and Characteristics

Mouth noises are sounds produced by movements of the lips, tongue, or mouth during speech. They can include clicks, smacks, lip smacks, breathing sounds, and even the subtle separation of the tongue from the palate. These noises are often unintentionally captured during recording and can be distracting if not addressed during editing. They originate from several sources:

  • Dry Mouth or Dehydration: A lack of saliva increases friction between the tongue and the roof of the mouth, producing sharp clicks.
  • Mic Placement: A microphone positioned too close to the mouth picks up exaggerated oral sounds that would otherwise go unnoticed.
  • Poor Recording Technique: Lack of pop filters or windscreens contributes to plosives and breath noises that can be mistaken for mouth sounds.
  • Natural Variation: Some speakers are more “mouthy” than others — certain vocal habits produce more clicks, especially after extended speech.
  • Background Silence: When the recording environment is very quiet, mouth noises become more prominent relative to the ambient noise floor.

Understanding these causes helps editors decide whether to address the issue during recording, through pre-processing, or via manual editing. However, for already-recorded material, the focus must be on removal techniques that preserve dialogue flow.

Core Strategies for Removing Mouth Noises

Editors have multiple approaches at their disposal. The best results often come from combining several methods, used judiciously. Below are the most reliable strategies, ranked from least to most aggressive in terms of potential impact on naturalness.

1. Spectral Editing Tools

Description: Spectral editing software, such as iZotope RX or Adobe Audition’s Spectral Displays, allows editors to visually identify mouth noises based on their unique frequency signature. A typical mouth click appears as a short, vertical line (wideband energy) in the spectrogram, often between 3 kHz and 8 kHz.

How to Apply: Zoom in on the waveform and spectrogram. Select the click with a rectangular or lasso tool, then apply a spectral repair (often with “Attenuate” or “Replace” mode). The algorithm fills the gap using surrounding frequencies, resulting in a seamless removal. For best results, use a short selection (2–10 ms). Overlapping selections can smear transients and dull the speech.

Pros: Highly precise; can remove clicks with almost no audible change in timbre. Works in context.

Cons: Requires a trained eye and ear; can be time-consuming if many clicks are present. Not all DAWs have built-in spectral editing – often an external plugin or dedicated application is needed.

2. Manual Cut and Fade

Description: For isolated, loud mouth noises that are not overlapped by speech (e.g., a distinct lip smack in a pause), a manual cut is simple and effective.

How to Apply: Place two edit points around the noise – just before and just after the sound – and cut the segment. Then apply a very short fade (1–5 ms) on both edges of the cut region to avoid a click from the edit itself. The result is a seamless removal of the unwanted sound.

Pros: Works in any DAW; no specialized plugins required. Quick for isolated noises.

Cons: Ineffective if the noise falls within continuous speech where cutting would create a noticeable skip or unnatural jump in the waveform. Also, if the cut is too wide, it removes part of the intended phoneme and distorts the word.

3. De‑noising and De‑clicking Plugins

Description: Many audio restoration plugins include de‑click or de‑noise modules specifically designed to target impulsive sounds. iZotope RX’s Mouth De‑click is a prime example; it uses machine learning to detect and attenuate mouth noises in real time or batch processing. Others include Accusonus ERA De‑click or Waves WLM Mouth De‑click.

How to Apply: Insert the plugin on the track and adjust threshold, sensitivity, and strength controls. Start with a low strength and listen critically. The goal is to reduce the clicks, not eliminate them entirely – over‑processing results in a “swishy” or “watery” artifact.

Pros: Fast, can process entire clips at once. Reduces many small noises that would take hours to edit manually.

Cons: Can introduce audible artifacts if overdone. May affect sibilance (S and T sounds) and high‑frequency clarity. Not a substitute for careful manual editing on problematic spots.

4. Automation and Keyframes (Volume Ride)

Description: Instead of removing the noise, you can simply lower its volume during the moment it occurs, using volume automation or clip gain keyframes.

How to Apply: Create a very narrow volume dip (roughly 10–20 dB reduction) at the exact moment of the mouth noise. The dip should be only as wide as the noise itself (5–10 ms). Use cross‑fade points to smooth the transition in and out. The surrounding dialogue stays untouched.

Pros: Preserves the original waveform; no destructive edits. Can be undone easily.

Cons: If the noise is very loud, a volume dip may leave a silent spot that still feels unnatural. If the dip is too wide, the word will sound muffled. This method is best for very brief, low‑to‑moderate clicks within sustained vowels or fricatives.

5. Equalization (Targeted Notch Filtering)

Description: Some mouth noises occupy a very narrow frequency band. A dynamic EQ or a static notch filter can reduce that specific range during the noise.

Use Caution: This is rarely a first‑line method because mouth noises have broad frequency content, and notching too much can also remove important voice harmonics. However, for a persistent high‑frequency whistle or a specific click that appears at a consistent pitch, a narrow Q filter can soften it.

How to Apply: Identify the offending frequency (use the spectrogram or a spectrum analyzer). Apply a dynamic EQ with a side‑chain set to trigger only when the noise is present. This preserves the dialogue when the noise isn't active.

Maintaining Dialogue Flow: The Editor’s Balancing Act

While removing mouth noises, it’s crucial to preserve the natural rhythm of speech. Over‑editing can make dialogue sound unnatural or choppy. To avoid this, consider the following advanced tips:

Edit in Small Sections

Work in increments of 5–15 seconds rather than attempting to fix an entire 30‑minute interview in one pass. Listen to the before and after of each small region. This allows you to catch subtle artifacts you would miss in a full‑length listen.

Use Crossfades Generously

Every cut, spectral repair, or volume automation point should have a very short fade or crossfade (1–5 ms). Without these, the edit itself can introduce a click that may be as noticeable as the original mouth noise. Crossfades also mask small timing discrepancies if your edits aren’t perfectly aligned with the waveform.

Listen in Context

Always review edits within the full dialogue to ensure natural flow is preserved. Solo‑ing a single word or phrase can trick your ear into thinking a removal is seamless, but when you hit play from a few seconds earlier, the transition may feel abrupt. Listen to at least two seconds of context before and after each edit.

Don’t Remove Everything

Natural dialogue includes tiny breaths, lip smacks, and other sounds. Removing every single one can result in a sterile, robotic quality. Aim to remove only those that are distracting or that stand out. Let some ambient mouth sounds remain, especially when they occur during pauses or between sentences. This maintains a human feel.

Advanced Techniques and Workflow Integration

Batch Processing with Spectral Repair

For projects with consistent mouth noise issues (e.g., a podcast recorded in one session), you can create a pre‑set in iZotope RX that applies a standard spectral repair to all clicks. Then, run the batch processor with a low attenuation setting. After batch processing, go back and manually fix any remaining loud clicks that the algorithm missed or over‑processed.

Using a Dedicated Mouth De‑click Plugin as a First Pass

Set up a mouth de‑click plugin on the track with a moderate setting. Render a draft track. Then, manually inspect the rendered audio, and use spectral editing only on the worst remaining noises. This workflow combines speed with precision. For example, iZotope RX offers a standalone application that integrates with most DAWs.

Working with Directus and Audio Content

If you are managing audio files through a headless CMS like Directus, ensure your asset pipeline includes a pre‑processing step. Upload high‑quality WAV files, then run a script or cloud function that applies a light mouth de‑click filter before the file is converted to a streaming format. This front‑loading reduces the manual editing workload for your editorial team downstream. For more on audio asset management, see Directus Asset Management Guide.

Prevention: Reducing Mouth Noises at the Source

No amount of editing can fully replace a clean recording. Implementing these recording‑side tips will drastically cut down on post‑production effort:

  • Hydrate the Talent: Encourage speakers to drink water (room temperature, not iced) before and during recording. A well‑hydrated mouth produces fewer clicks.
  • Mic Technique: Use a cardioid or hypercardioid microphone placed about 6–12 inches from the mouth, slightly off‑axis (not directly in front). This reduces plosives and sibilance while also softening mouth noises.
  • Pop Filters and Windscreens: A pop filter catches bursts of air that cause plosive pops. A foam windscreen can further diffuse breath and lip sounds. For indoor dialogue, a foam screen is often sufficient.
  • Room Treatment: A dead‑sounding room (with acoustic panels) will have less room tone, making mouth noises more prominent? Actually, a treated room reduces reflections that can mask mouth noises? No – the opposite: a dead room actually highlights mouth noises because there is less ambience. However, it also reduces reverb that can smear clicks. The ideal is to have a controlled but not completely dead space. Use a small amount of broadband absorption to avoid hollow sounds.
  • Pre‑recording Vocal Warm‑ups: Simple humming, lip trills, and tongue exercises moisten the mouth and loosen the vocal apparatus, reducing the frequency of clicks.

Common Pitfalls to Avoid

  • Over‑Processing: Applying heavy de‑click processing to an entire track can cause a “swirly” or “metallic” artifact. Always use bypass comparison and check with headphones.
  • Removing Breaths: While some breaths are noisy, they are often necessary for natural pacing. Only remove breaths that are loud or gasping. Use fades to blend the remaining breath instead of deleting it entirely.
  • Ignoring Room Tone: After removing mouth noises by cutting or spectral repair, you may create gaps of absolute silence. Fill these with a short clip of room tone (recorded from the same space) to maintain a consistent background ambiance.
  • Rushing the Listen: Always test your edits on different playback systems (speakers, headphones, laptop speakers). What sounds clean on high‑end monitors may reveal clicks on earbuds.

External Resources for Further Learning

For editors who want to dive deeper into audio restoration and dialogue editing, the following resources are highly recommended:

Conclusion

Effective editing of mouth noises enhances the clarity and professionalism of your content, ensuring viewers stay engaged with the dialogue without distraction. By combining spectral editing, manual cuts, plugins, automation, and a healthy dose of critical listening, editors can achieve clean dialogue that sounds completely natural. Remember that the best strategy is often a hybrid approach: use automation or a de‑click plugin for the majority of small clicks, then zoom in with spectral repair for the few stubborn pops. And always, always listen in context. With practice, these techniques become second nature, allowing you to deliver polished audio that lets the story – not the sounds of the human mouth – take center stage.