Tools
Remove silence, pauses, and filler words from a video
Vyra finds pauses, ums, and repeated takes in the transcript and cuts them at a threshold you set. Small breaths stay so it still sounds like you.
By Sulan Zhang, co-founder of Vyra · Updated 2026-09-24
- Best for
- Talking head, tutorial, podcast, and course recordings
- Input
- Any video with speech
- Output
- MP4 up to 4K, or WebM up to 1080p
- Time
- Minutes for a 10-minute recording once processed (TODO confirm)
- Works with
- Built-in chat, Claude, ChatGPT, Cursor over MCP
- Price
- Free to start, plans from $24/mo
Cut the dead air, the ums, and the false starts so the recording sounds like the version you meant to say. Vyra does it from the transcript with a pause threshold you pick, and keeps the small breaths.
How it works
- Upload the recording. Transcription runs on its own.
- Tell the chat what to remove. Pauses over a threshold, filler words, retakes, or all three.
- The agent cuts those parts and closes the gaps.
- Ask for a list of cuts with timestamps and check any you're unsure about.
- Hide rough jump cuts with a punch-in or b-roll if you want, then export.
What you can control
| Setting | Options |
|---|---|
| Pause threshold | Any value. 0.3, 0.5, 1 second |
| Filler words | um, uh, like, you know, or your own list |
| Retakes | Keep the last attempt, the first, or one you name |
| Protected regions | Timestamps it must not touch |
| Cut treatment | Jump cut, punch-in on the second clip, or cover with b-roll |
| Range | Whole recording or a section |
Prompts to paste
Clean up this talking head. Remove ums, uhs, pauses longer than half a second, and any sentence I restart. Keep the best version of each sentence. Leave small natural pauses so it doesn't feel robotic.
Cut every pause over 0.4 seconds. Where a cut is jarring, punch in 10 percent on the second clip instead of a jump cut. Don't touch 1:20 to 1:35, that pause is on purpose.
Remove "like" only when it's filler, not when I mean it. Then list every cut with timestamps.
When to use something else
- You need the cleanup inside Premiere, Final Cut, or Resolve. Gling and similar plugins hand off to those.
- The recording is mostly music or ambient sound. This is a speech tool.
- Two people talking over each other the whole time. Separate the tracks first.
FAQ
Will it make me sound choppy? Not if you pick a sensible threshold and ask it to keep natural pauses. Half a second is a good start.
How do I check what it removed? Ask for the list of cuts with timestamps. Takes ten seconds to compare.
Can it remove one specific repeated line? Yes. Quote it. "Remove the repeated 'so basically' at 2:14."
Related
- /prompts/talking-head-cleanup
- /prompts/remove-filler
- /how-to/cut-a-talking-head
- /tools/edit-video-by-typing