Techniques for Reducing Mouth Noise and Breaths in Dialogue Tracks

Understanding Mouth Noise and Breaths in Dialogue

Unwanted mouth clicks, lip smacks, saliva bubbles, and audible breaths are among the most common imperfections in dialogue recordings. While the human ear naturally filters these sounds in real life, microphones capture them with startling clarity. In professional audio post‑production, removing or reducing these artifacts is critical to maintain immersion, clarity, and a polished final mix. This article covers proven techniques—from pre‑recording preparation to advanced editing workflows—so you can deliver dialogue that sounds clean, natural, and engaging.

Pre‑Recording Preparation to Minimize Noises

Microphone Technique and Placement

The first line of defense against mouth noise is smart microphone positioning. Position the microphone 6–12 inches from the speaker’s mouth, slightly off‑axis (pointing just to the side of the mouth instead of directly at it). This reduces the intensity of plosives and saliva pops while preserving vocal clarity. Use a pop filter—a fine mesh screen placed between the speaker and the microphone—to diffuse bursts of air before they hit the capsule. For voice‑over work, a windscreen (foam cover) on the microphone provides additional protection against breath noise, though it may slightly affect high‑frequency response.

Speaker Guidance and Hydration

Instruct the talent to control their breathing rhythm. Ask them to take quiet, deep breaths during pauses rather than sharp, shallow gasps. Encourage them to moisten their lips and drink room‑temperature water regularly; dry mouth exacerbates sticky sounds. Avoid dairy products and sugary drinks before a session, as they can increase saliva viscosity. Many voice actors also practice pre‑session warm‑up exercises—like humming or tongue twisters—to reduce oral tension and excessive saliva accumulation.

Acoustic Environment

Record in a treated space or use a reflection filter to reduce room echo. A dry, dead acoustic environment makes it easier to process breaths and mouth noises without competing with reverb. If you are capturing dialogue on location, place the microphone close while using a boom or stand to minimize handling noise and wind.

Post‑Production Workflows for Noise Reduction

Manual Waveform Editing

Zoom into the waveform at the waveform’s scale and visually identify breath or click artifacts. Use a razor tool to cut out isolated breath sounds, then apply a short crossfade (2–5 ms) to avoid clicks. For mouth clicks that fall within speech, use a spectral editing tool to reduce their amplitude without removing the underlying vocal. Izotope RX’s Spectral Repair, Adobe Audition’s Auto Heal Selection, and FabFilter Pro‑Q’s dynamic EQ can all be used to surgically remove these artifacts. Izotope’s guide to removing mouth clicking in RX offers a detailed workflow.

Automated Breath Reduction Tools

Modern DAWs and dedicated plugins include automatic breath‑detection algorithms. For example, Waves Renaissance DeBreath analyzes the waveform and reduces the volume of breathing segments based on a threshold. Accusonus ERA‑N (Noise Remover) uses machine learning to isolate and reduce mouth clicks, breaths, and rumble simultaneously. These tools are fast but may overprocess; always audition the result and adjust parameters like sensitivity and attack to maintain a natural flow. Waves’ DeBreath plugin details can help you understand proper settings.

De‑Essing and EQ for High‑Frequency Noises

Many mouth clicks and lip smacks reside in the 4–10 kHz range. Use a de‑esser (often included in EQ plugins) to reduce this frequency band only during noisy segments. A multiband compressor can also be effective: set a band with a low threshold in the 5–8 kHz range, with a fast attack (1–2 ms) and release (50–100 ms). This tames harsh clicks without dulling the overall vocal presence. Some engineers apply a gentle notch filter at the specific frequency of a recurring click—sweep through the spectrum while the noise is playing to identify the exact resonance.

Compression and Leveling for Breath Consistency

Breaths that are too prominent often result from uneven dynamics. A compressor with a low ratio (2:1 or 3:1) and a medium threshold can even out the level of breaths relative to the speech. Alternatively, use volume automation to manually duck breaths by 4–8 dB. Make the reduction gradual to avoid an unnatural sudden volume drop. Some post‑production houses keep breaths at around –20 dB relative to the dialogue peak to maintain realism while preventing distraction.

Advanced Spectral Editing and Restoration

Using Spectral Layers for Click Removal

For stubborn click artifacts that cut through the dialogue, spectral editing is the gold standard. In iZotope RX or Sound Forge, switch to spectral view and select the click region. Use the “Cure” or “Replace” function to reconstruct the missing audio using surrounding frequencies. This method preserves the natural timbre while eliminating the pop. For mouth noises that occur on sibilant consonants, use a dedicated de‑click module that analyzes the transient shape.

Batch Processing Multiple Takes

When cleaning a long-form project (podcast, audiobook, film), create a processing chain in your DAW or use a standalone audio repair tool that can batch process similar segments. For example, iZotope RX’s Batch Processor lets you apply a set of modules—de‑click, de‑breath, de‑ess, and spectral repair—to every file in a folder. Save custom presets to maintain consistency. Always verify a few processed samples to confirm no unnatural artifacts were introduced.

Practical Workflow for Dialogue Editors

Step‑by‑Step Guide

Prepare the recording: Ensure proper mic placement, pop filter, and hydrated talent.
First pass – manual cuts: Remove obvious breath silences between phrases (keeping natural breath at pause points).
Second pass – spectral repair: Zoom into waveform and surgically remove clicks and mouth pops using spectral editing.
Apply de‑breath plugin: Use a plugin like Waves DeBreath or Accusonus ERA‑N to reduce remaining breaths by 6–10 dB.
De‑ess and EQ: Apply a de‑esser targeting 5–8 kHz, then a gentle high‑shelf cut (–2 dB at 10 kHz) to reduce sibilance and mouth fricatives.
Compress and level: Use a light compressor (2:1 ratio, –20 dB threshold) to even out loudness, then automate breath volume if needed.
Final listen: Play back in context with background sounds (music, SFX) to ensure breaths are not distracting.

This workflow typically reduces mouth noise by 80–90% while keeping the dialogue sounding natural. For extreme cases, consider re‑recording the worst lines rather than overprocessing.

Tools and Plugins Comparison

Common Mistakes and How to Avoid Them

Over‑Processing That Creates Unnatural Sound

Aggressive removal of breaths can make dialogue sound sterile or robotic. Always leave some natural breaths—especially at emotional moments—to maintain realism. Use a de‑breath plugin with a low threshold (e.g., reduce by 4–6 dB) rather than completely muting breaths. Similarly, excessive spectral editing can introduce clicks or watery artifacts; use the “Listen” button to preview repairs before applying.

Ignoring Room Tone and Background Noise

Cutting out breaths can leave awkward silences where room tone (ambient noise) is missing. After removing a breath, paste in a short snippet of room tone from the same recording to fill the gap. This prevents the track from sounding hollow. Many editors keep a clip of pure room noise from the session and patch it over deleted segments.

Final Best Practices for Clean Dialogue

Record multiple takes and select the one with the least mouth noise—editing cannot replace a good performance.
Use a consistent microphone distance and pop filter across all sessions to minimize variability.
Monitor your processed track through headphones and speakers; what sounds clean on one system may pop on another.
Save presets for your processing chain so you can apply them instantly to subsequent projects.
When in doubt, err on the side of naturalness—slight audible breaths are preferable to artificial, hollow dialogue.

By combining preventative recording habits with careful post‑production editing, you can achieve dialogue tracks that are clear, professional, and free from distracting mouth noise. The techniques outlined here are used by audio engineers in film, television, podcasting, and audiobook production and scale to any project size.