Editing by Deleting Words

Editing by Deleting Words

Using the transcript as the timeline, and the two edits that produce an audible seam.

descript video audio editing

What it is for

Editing by deleting words is fast and produces two audible seams if you are not watching for them: a cut across a breath, and a cut that changes the room tone. This covers the workflow and both seams.

The skill

The transcript IS the edit. Work there and only drop to the waveform for the
seams.

The pass that does most of the work, in order:
1. Run filler-word removal, then **read the result** rather than accepting it.
   It removes "like" and "you know" as fillers even when they are the sentence.
2. Delete whole sentences, not phrases. A half-deleted sentence changes the
   speaker's pitch contour mid-breath and that is what the ear catches.
3. Cut on a BREATH, not on the word boundary the transcript shows. The
   transcript's boundary is where the word ends; the natural cut is a beat
   later, and the difference is the whole seam.

Two edits that always sound wrong:
- **Joining two takes at different distances from the microphone.** No amount
  of levelling fixes the room tone change. Re-record the line.
- **Removing a pause entirely.** A speaker who never pauses sounds anxious.
  Shorten pauses to about 60% instead of deleting them.

Use Overdub only for single words you got wrong, never for a new sentence. A
generated sentence carries no breath and sits oddly between two real ones.

How to install it

1. Open the tool and start a new conversation, project or template.
2. Paste the text below wherever the tool keeps standing instructions
   — usually called system prompt, instructions, or persona.
3. Save it so it applies to every run rather than a single message.

Did it work for you?

Where it came from

Skills app Original