Remove filler words from a video on your Mac
Yaptake removes filler words from a video on your Mac, without uploading it. It transcribes the recording locally, so each “um” and “uh” is a word in the transcript: delete it and its sound and picture go too, or ask the built-in chat to cut them all in one step. There are no credits or minutes, and the original file is never changed.
Four ways to cut filler
- Delete the words. Double-click an “um” in the transcript and press Delete. The sound and picture leave the cut together. ⌘F finds each one in the transcript.
- Ask the chat. Press ⌘K and type “Cut the ums, uhs and long pauses”. The chat runs Gemma 4 on your Mac. The whole request is one undo step, and a request that would cut more than a quarter of the video waits for you to press Apply.
- Make a rough reel. Edit → Make Rough Reel from Library trims each library video to its speech, keeps the last complete take of a line said twice, and cuts filler words and measured silence, as one undo step.
- Hand it to a coding agent. Claude Code or Codex can run
cut_fillers()through Yaptake’s command line or MCP server on the project open in the app. The walkthrough shows the setup.
Which words count as filler
Hesitation sounds: um, uh, er, erm, hmm, mm, ah and eh, and longer spellings such as umm, uhm and uhh. Words such as “like”, “so” and “you know” are left alone, because speakers often mean them. Find them with ⌘F and delete the ones you want gone.
Pauses and dead air
Edit → Tighten Pauses shortens every pause longer than 0.4 seconds to 0.4 seconds, in the selection or in the whole video. With Edit → Include Silence checked, it also removes silence measured in the sound inside those pauses, keeping 0.1 seconds of quiet next to speech. Each run is one undo step.
A hesitation the speech model did not write down sits inside a pause, so tightening pauses shortens it too.
Nothing is lost
Cut words stay in the transcript, struck through. Select them and choose Edit → Restore Deleted Words to bring them back, or use Undo. Yaptake never rewrites your recording; the project only records what to leave out.
Why on the Mac
Speech-to-text and the chat model run on the Mac’s own chip. Cleaning up speech needs no upload, no account and no AI credits, and it works with Wi-Fi off once the models are on your Mac. The speech model is about 632 MB and downloads in the background on first launch. The chat model is about 5 GB, or about 17 GB on a Mac with 32 GB of memory or more, and downloads only when you choose Download.
Limits worth knowing
- English speech only.
- Apple Silicon Macs on macOS 26 or later.
- Nothing is cut when you import. Filler goes only when you delete it or ask.
- Exports are 1080p at 30 fps in 9:16, 4:5, 1:1 or 16:9.
Questions
Can I remove filler words without uploading my video?
Yes. Yaptake transcribes and cuts on your Mac. The video, its audio and its transcript stay there.
Does it work offline?
Yes, once the speech model has downloaded, and the chat model if you want to ask the chat. Deleting words and tightening pauses need no chat model.
Does removing filler words cost credits?
No. There are no credits or minutes. Editing is always free. Your first 3 successful exports on each Mac are free. After that, exporting is $9.99 a month or $79 a year.
Can I undo a filler cut?
Yes. Each cut, chat request or rough reel is one undo step, and cut words can be restored from the transcript at any time.
Does Yaptake cut “like” and “you know”?
Not by itself. Those words often carry meaning, so they are left for you to find and delete.
Try Yaptake
Your first 3 successful exports on each Mac are free, with no watermark, account or credit card required. After that, exporting is $9.99 a month or $79 a year. Editing is always free.
