VSE Audio Cleaner — Documentation 

Installation

1. Download the VSE Audio Cleaner .zip file
2. In Blender, go to Edit > Preferences > Get Extensions
3. Click the dropdown arrow near the top right and choose Install from Disk
4. Select the .zip file — it installs as "VSE Audio Cleaner"
5. Confirm the checkbox next to it is enabled

Requires Blender 4.2 or newer. No external dependencies, no internet connection needed.

Where to find it

Open the Video Editing workspace, press N in the Sequencer to open the sidebar, and select the Audio Cleaner tab. You'll see the Silence Cleanup panel with three collapsible sections below it: Normalize, Duck Music Under Speech, and Auto-Sync.

A note before you start

All four tools work on sound strips only. If you added your video using Add → Movie with "Sound" enabled (the default), Blender already creates a separate sound strip for you automatically — you don't need to do anything extra. If your video strip has no matching sound strip, split the audio out first: Strip → Audio → Extract, or re-add the video via Add → Movie with Sound checked.

Feature: Silence Cleanup

What it does: Scans sound strips for silence and dead air, lists every range it finds, and lets you either delete or quiet each one.

Workflow:

1. Select the sound strip(s) you want to scan (or disable "Selected Strips Only" to scan every sound strip in the project)
2. Adjust settings if needed (defaults work well for typical spoken dialogue)
3. Click Detect Silence
4. Review the list — uncheck anything you don't want touched
5. Choose an Action: Delete (Ripple) or Attenuate Only
6. Click Process Checked Ranges

Settings explained:

• Silence Threshold (dB) — audio quieter than this counts as silence. More negative (e.g. -50) is stricter and only catches near-total silence. Less negative (e.g. -30) also catches quiet room tone.
• Minimum Duration (sec) — only flags gaps at least this long, so natural pauses between words aren't caught unless you want them to be.
• Padding (sec) — shrinks each detected range inward on both ends, so a cut never lands right on the edge of a word.
• Action → Delete (Ripple) — cuts the range out entirely and shifts everything after it left, shortening your timeline.
• Action → Attenuate Only — keeps every cut exactly where it is and just lowers the volume through the range. Use this for room tone or background noise you want quieter, not gone.
• Attenuate Amount (dB) / Fade Time (sec) — how much quieter, and how smooth the transition is (only shown in Attenuate mode).

Feature: Normalize

What it does: Measures each strip's level and adjusts its volume so it matches a target — useful when strips were recorded on different microphones or in different rooms.

Workflow:

1. Select the strip(s) to normalize (or disable "Selected Strips Only")
2. Choose a Reference mode and Target level
3. Click Normalize Levels

Settings explained:

• Reference: Peak — normalizes so the single loudest sample in the file hits your target. Good for avoiding clipping.
• Reference: RMS (loudness) — normalizes overall perceived loudness. Usually the better choice for spoken dialogue.
• Target Level (dB) — the level you want strips to match. -14 dB RMS is a reasonable starting point for voice.
• Max Boost (dB) — caps how much a quiet file can be amplified, so a near-silent recording doesn't get boosted into audible hiss.

Feature: Duck Music Under Speech

What it does: Automatically lowers a music strip's volume wherever dialogue is active, and brings it back up in the gaps.

Workflow:

1. In the Sequencer, select only the music strip(s) you want ducked
2. Make sure every other sound strip is deselected — they're all treated as dialogue
3. Open the Duck sub-panel, adjust settings if needed
4. Click Duck Selected Music Under Dialogue

Settings explained:

• Voice Threshold (dB) — audio above this level on the dialogue strip(s) counts as active speech.
• Minimum Phrase (sec) — ignores very short speech blips shorter than this.
• Bridge Gaps Under (sec) — treats short pauses between words/sentences as still-speaking, so the music doesn't pop back up between every word. Increase this if the music flickers too much; decrease it if music recovers too slowly during real pauses.
• Duck Amount (dB) — how much quieter the music gets while speech is active.
• Fade Time (sec) — how long the volume ramp takes going down and coming back up.

Note: Running Duck and Attenuate on the same strip will replace the first one's volume automation, not combine with it. Use one or the other per strip.

Feature: Auto-Sync

What it does: Aligns two separately recorded audio strips (e.g. a camera mic and a separate recorder capturing the same moment) using waveform cross-correlation — no clapperboard required.

Workflow:

1. Select exactly two sound strips
2. Open the Auto-Sync sub-panel
3. Click Auto-Sync Selected Strips
4. The later strip moves automatically; the reported offset shows in the panel

Settings explained:

• Analysis Window (sec) — how much of the start of each strip is compared to find the match. If the two recordings don't share an obvious audio event (a clap, a door, matching room tone) within this window, widen it.

Note: Auto-Sync needs the two recordings to actually overlap somewhere in the analyzed window. It can't align two clips that share no common audio event.

Tips & Best Practices

• Tune your Silence Cleanup threshold once for your mic and recording space — it'll hold for future takes with the same setup.
• Run Silence Cleanup and Normalize before Duck, so the ducking analysis is working with already-cleaned dialogue.
• For Auto-Sync, a single clap or door slam at the very start of a recording session makes alignment fast and reliable — build that into your recording habit if you regularly record dual-source audio.
• Every action here is a single Ctrl+Z step, so don't hesitate to experiment with settings and undo if a result isn't what you wanted.