Detect Umms

Detect Umms finds filler sounds such as "um," "uh," "er," and "ah." You can save the detections for review or remove the fillers from the timeline right away.

Before you run it

Detect Umms needs a transcript for each recording it scans. Transcribe your footage first. Media without a transcript is skipped.

Run Detect Umms

1. Open the Process menu and click Detect Umms….
2. Under Tracks, check the audio tracks to scan. Tracks with nothing transcribed are greyed out. If only one track has transcribed media, it starts checked.
3. Add any words you want treated as fillers under Extra filler words, separated by commas or spaces. For example, enter "kinda, sorta."
4. Under After detection, choose what should happen to the timeline.
5. Click Run.

Your settings are remembered for the next run, except the track selection. For cancelling a run or handling a failure, see Run an AI Tool.

Choose what happens after detection

• Detect Only saves the results without changing the timeline. This is the default, and it's the safer choice when you want to review the cuts first.
• Disable splits each filler from the surrounding clip and turns that piece off. Playback skips it, but nothing moves. You can turn the pieces back on or remove them later. See Cut and Delete.
• Delete removes each filler and leaves a gap.
• Ripple Delete removes each filler and closes the gap by moving later clips left.

Any choice except Detect Only applies the result when detection finishes. It only applies to the tracks you selected. Linked audio and video are cut together, so you can't remove one without the other.

For Disable, Delete, or Ripple Delete, you can turn on Add audio fades at cuts. It is off by default. When a cut lands in speech, this adds a short fade of about 20 ms to soften clicks at the splice. Tiny cuts with no clean splice point are skipped instead of forced.

Adjust gaps and pauses

Open Advanced if the default cuts feel too tight or too loose.

• Min clear gap sets how much silence must sit on both sides of a filler before it can be removed. It runs from 50 to 500 ms and defaults to 150 ms. Raise it to keep fillers that sit close to other words.
• Remaining pause sets how much pause stays after the filler and nearby silence are removed. It runs from 0 to 1000 ms and defaults to 250 ms. Raise it for more breathing room, or lower it for tighter pacing.

Apply saved detections later

If you chose Detect Only, you can apply those results later:

1. Select the audio clips you want to clean up.
2. Open the Clip menu and click Remove Umms….
3. Choose Disable, Delete, or Ripple Delete. Delete is the default.
4. Turn on Add audio fades at cuts if you want short fades at speech seams.
5. Click Apply.

Only selected clips with saved Detect Umms results are changed. Run Detect Umms on any remaining clips before applying if you want them included. Linked audio and video are cut together.

What happens to the transcript

Detect Umms never rewrites the transcript. Detected words stay in the transcript even when their ranges are disabled or removed from the timeline. Fillers without enough clear space are kept.

Skipped cuts and undo

With Add audio fades at cuts turned on, a cut that is too small to make cleanly is skipped and marked on the timeline as Cut skipped: no clean splice point. See check for skipped cuts.

Whether you apply the cuts during detection or later with Remove Umms…, one press of Ctrl+Z undoes the whole pass, including fades and skipped-cut markers.