Dialogue Editing & Phase Alignment: Lav vs Boom Guide

1. The Physics of Phase Smearing: Why Lavs & Booms Clash

Sound travels through air at approximately 343 meters per second (roughly 1 foot per millisecond):

• The Time Delay Offset: A lavalier microphone sits directly on an actor's chest (6 to 8 inches from their mouth), while an overhead boom microphone typically hovers 2 to 3 feet above. The sound wave hits the lavalier capsule approximately 2 to 3 milliseconds before it reaches the boom mic.
• Comb Filtering Destruction: When you sum both signals to a single mono dialogue bus without time-alignment, frequencies where the signals are 180 degrees out of phase cancel out completely. This creates a series of deep notches across the vocal spectrum (comb filtering), making the actor sound hollow, thin, and nasal.

2. The Primary Golden Rule: Choose One Mic First

The most common mistake amateur dialogue editors make is leaving every recorded microphone open simultaneously:

3. Manual Sample-Accurate Phase Alignment

When an artistic scene demands blending both the close-mic punch of a lavalier and the acoustic air of a boom:

1. Zoom in to the Sample Level: In your DAW (Pro Tools, Nuendo, or Reaper), zoom in on a loud plosive consonant (like a "P" or "B" sound).
2. Identify the Delay Offset: You will see the lavalier waveform spike occur several hundred samples before the boom waveform.
3. Nudge the Boom Track: Slide the boom audio region backward in time (or insert a sample-delay plugin) until the initial compression peaks align with zero phase offset.
4. Check Polarity: Flip the phase (180° polarity switch) on the lavalier channel to verify which orientation yields the fullest bass response and solid dialogue presence.

4. Dynamic Phase Alignment with Dedicated DSP Plugins

Because an actor moves around during a scene, the physical distance between their mouth and the boom pole constantly shifts:

Static time-delay adjustments quickly drift out of alignment as the boom operator swings the pole. Professional post-production facilities rely on dynamic spectral alignment tools such as Sound Radix Auto-Align Post 2 or MeldaProduction MAutoAlign. These plugins continuously analyze the time-varying delay and apply micro-pitch phase tracking across the entire scene automatically.

5. Track Checkerboarding & Room Tone Smoothing

Dialogue editing is the art of invisible transitions:

• Checkerboarding: Separate dialogue lines by character onto distinct DAW audio tracks (Character A on Track DX 1/2, Character B on Track DX 3/4). This allows tailored EQ, compression, and noise isolation without affecting the other actor's vocal tone.
• Equal-Power Crossfades: Use 5 to 10 ms equal-power crossfades at dialogue edit points to eliminate digital clicks and pops.
• Room Tone Fills: Place a continuous bed of matching location room tone across track cuts so the background noise floor never drops to dead digital silence.

Measured: where the comb filter notches fall for a lav and boom offset

We summed a signal with a delayed copy of itself on 2026-10-01 to see the comb filter section 1 describes. The signal was six seconds of white noise, the copy was delayed in whole samples at 48 kHz with FFmpeg 8.1.2 (adelay, then amix with normalising off) and set 6 dB lower, and the notches were read from the measured response. Nothing was recorded, so this shows what a given offset does and not what a particular lav and boom do. The theory predicts a first notch at 1 divided by twice the delay, a notch every 1 divided by the delay, and, for a copy at level ratio r, notches 20 log10(1 - r) below the original and peaks 20 log10(1 + r) above it. For r = 0.5 that is 6.0 dB down and 3.5 dB up, a swing of 9.5 dB.

The measured notches sit where the theory puts them, within the 2.9 Hz resolution of the analysis. A 2 ms offset puts the first notch at 250 Hz and one every 500 Hz above it, so there are 16 notches between 250 Hz and 8 kHz, which is the hollow sound section 1 describes. The notches are only 6 dB below the original here, with 3.5 dB peaks between them, because the copy is 6 dB down. With a copy at equal level the same test gave dips deeper than 30 dB, limited only by the analysis resolution, which is what "cancel out completely" would need, and a lav and a boom are rarely level matched.

The page's "several hundred samples" does not match its own numbers. 2 to 3 ms is 96 to 144 samples at 48 kHz, so several hundred would be 6 to 10 ms or about 2 to 3.5 m of extra path. The distances in section 1 give less than 2 to 3 ms as well. A boom 2 to 3 ft above and a lav 6 to 8 in from the mouth differ by 16 to 30 in of path, 0.41 to 0.76 m, which is 1.2 to 2.2 ms if the boom is straight above, the lav is on the same line and the speed of sound is 343 m/s. The real offset depends on the angle and moves with the actor, which is why step 3 says to zoom in and measure it.

Alignment only has to be close. A 12 sample miss (0.25 ms) puts the first notch at 2000 Hz and a 3 sample miss puts it at 8000 Hz, so the damage of a small residual error is pushed up the spectrum, but whether it is audible was not tested. The test also says nothing about how well a plugin tracks a moving boom.

Validated 2026-10-01 on white noise with FFmpeg 8.1.2. Delays were whole samples at 48 kHz and no microphone, actor or alignment tool was used. The geometry is the page's distances in a straight line.

Working Vocabulary

Key dialogue editing and acoustic phase terms:

• Comb Filtering. An acoustic interference artifact where summing delayed signals creates periodic cancellation notches across the frequency response.
• Phase Alignment. Adjusting the time relationship between two or more audio tracks to align their waveforms at the sample level.
• Checkerboarding. Organizing dialogue clips onto dedicated, alternating DAW tracks categorized by actor and shot angle.
• Sample Delay. A micro-timing adjustment (measured in samples or fractions of milliseconds) used to eliminate phase offsets.
• Polarity Inversion. Flipping an audio signal's electrical phase by 180 degrees (inverting positive and negative voltage swings).
• Spectral Repair. Surgical audio restoration using frequency-domain spectrogram editing to remove clicks, cloth rustle, and mic bumps.

Sources & Further Reading

• What Happens When We Don't Need Room Tone Anymore?
• Boom Mic Operating & Location Sound Recording Guide
• Vocal Compression: Opto, FET, VCA & Tube Topologies
• Audio Preamp Gain Staging & SNR Calculator