All templates

A script for any video

Clean edges at every cut

by Takehound

Finds where you get ready or wind down at the start, the end and every cut, takes it off where it falls in silence, and lists the rest. Never clips a word.

For any video, 17 steps

Try it on my footageSign in with Google for a free first video

What Hound does with it

  1. Goal: no cut shows the person getting ready or winding down. Around the words, the picture can show them settling in, glancing down at notes, dropping their hands, leaning toward the camera or reaching to stop it. A viewer sees that even when the sound is clean, and it reads as an unedited cut. Never clip a word to hide it.
  2. Read their notes, then the transcript and the timeline. List the first word and the last word of the talk with their seconds, and every join where the talk was cut, with the last word before it and the first word after it. Leave alone every cut they made by hand.
  3. Watch the whole video in order, in stretches of 10 minutes at most. Ask the same of each: In the second before and after each cut, and before the first word and after the last, does the speaker look away from the lens, look down, drop or raise their hands, shift in their seat, lean in, or reach toward the camera? Give the second of each, and say whether they are speaking or silent then.
  4. Look yourself, with stills, at every moment the watch names, and keep only what you see. A beat held while they still look into the lens, before a number or a punchline, is delivery and stays.
  5. Take silence off only with tighten_pauses, never with cut_time at the transcript's times: the transcript sometimes gives a pause to the word beside it, and a cut at its times clips the word. Once the talk's sound has been read, tighten_pauses finds the pause by the sound and keeps the word whole. Use the transcript's seconds only to choose its from and to.
  6. The line The talk's sound in project/timeline says whether it has been read. When it says not read yet, cut nothing at the start, the end or any join: tighten_pauses would cut by the transcript's times too. List each with its second and what you saw, and say the sound was not read yet.
  7. Any cut takes time from every track. Before each one, read the timeline at that moment: if a title, a card, a clip or music sits there and would be shortened or moved, list it instead of cutting.
  8. The start: if they settle in before the first word, tighten_pauses with from 0, to the first word's second, longer_than 0.4 and leave 0.2. If a title or card plays at the start, list it instead of cutting. If they are already moving when the first word begins, change nothing there and list it.
  9. The end: if after the last word they look away, drop their hands or reach for the camera, tighten_pauses with from the last word's second, to the end of the talk, longer_than 0.4 and leave 0.2. If that would shorten a card, a clip or music placed after the talk, list it instead of cutting.
  10. A join where the movement falls in the silence between two words: tighten_pauses with from the end of the word before, to the start of the word after, longer_than 0.4 and leave 0.2, so only that pause shortens. If a card, a clip or music sits across the join, list it instead of cutting. Never take from inside a word, and never join two sentences with no breath between them.
  11. A join where the movement falls while they are speaking cannot be taken off without clipping a word. Change nothing there. List it with its second, what you saw, and what could fix it: another take of the line if they said it again (read the transcript for one), or one of their own clips over those seconds.
  12. Never cover a join with stock, with a clip of anyone else, or with a card that says what was not said. Only their own footage, and only on their yes.
  13. If nothing shows them getting ready or winding down, say so in one line and stop.
  14. Then stop and ask about the joins on the list: name one to cover with a clip of theirs, to swap for another take, or to leave as it is. Change nothing more on your own.
  15. When they answer, do only what they named, each as one step to undo: place_shot with one of their own clips over the seconds named, or cut_words on the take they no longer want and restore_words on the one they choose.
  16. Check: watch two seconds either side of every place you changed. No word is clipped or doubled, the first and last words play whole, no join shows them moving, and no card or clip lost its place. Fix only what you saw. If a fix of yours is undone twice for the same reason, stop and ask.
  17. Say how many cuts you looked at, what you took off and where with the seconds each lost, the talk's length before and after, and under Worth a look every join you left with its moment and what you saw. Say what this pass cannot judge: a small movement between two stills can be missed, so the joins on the list are worth their own look.

How it works

  1. Sign in with GoogleYour first video is free, a recording of up to 30 minutes.
  2. Open the editor on this scriptAdd your talk and your own footage. The editor opens on this template.
  3. Watch Hound workStop him at any point. Every change he makes can be undone.

More scripts

All scripts

A free first video

Run any template on your own footage

Sign in with Google and the editor opens, with every template here. Your first video is free, a recording of up to 30 minutes. More videos use credits, and they never expire.

Start free with Google

No card, no install. Privacy