Removing 'um'/'uh': clean your audio automatically
You don't need a perfect take
Every screencast has them — "um", "uh", "er", "like" (when you mean nothing), "you know". They slip in while you're thinking, and they survive into the final video because editing them out manually is tedious.
TutDub handles this automatically.
What happens under the hood
When you submit a video, the cleaning stage runs a language model over the transcript before any dubbing happens. It identifies filler words and disfluencies — "um", "uh", "er", "hmm", "ah", "like" used as filler, and repeated hesitation markers — and strips them from the text that gets synthesized into speech.
You don't toggle this on. It runs on every job by default.
There's also a noise-removal pass (on by default) that cleans up mic background noise — fan hum, keyboard clatter, room tone — from the audio track while leaving your voice untouched.
Together, the two passes give you dubbed audio that sounds cleaner than most raw screen recordings.
If you want filler removal without dubbing into another language, just pick your own source language as the target. You'll get the cleaned audio back in the same language — useful for polishing a raw recording before you publish it.
Zero extra effort
Neither feature costs extra credits or requires a setting. Upload your video, pick your languages, and the cleaning happens as part of the standard processing pipeline — right between transcription and voice cloning. Pricing is simple: 1 credit = 1 minute of video.
Your viewers hear clean, natural speech. You hear nothing you'd want to edit.
What's next
Next time we'll look at how TutDub lets you translate your video into 14 languages — all in your own voice, from a single upload.
Read next
- Translate your video into 14 languages, in your voice
- Stop being scared how your voice sounds — voice cloning, reassured
FAQ
-
Does TutDub remove filler words from every language?
Yes. The filler-removal step works on the transcript regardless of source language. It catches common hesitation markers in whatever language you upload.
-
Can I turn off filler removal?
Not currently — it's always on. There's no reason to keep "um" in a professional dub.
-
What about "um" in languages other than English?
The model recognises filler patterns across supported languages, not just English "um"/"uh". Each language has its own common hesitation markers, and the LLM is prompted to catch them.
-
Does noise removal affect my voice?
No. The denoiser runs on the separated background audio track, not on your speech. Your voice stays as-is; only the background gets cleaned up.
-
Do I need to enable these features manually?
No. Both filler removal and noise removal are on by default for every job. You can disable noise removal in the upload options if you want to keep background audio intact, but filler removal is always active.
-
Does cleaning add extra processing time?
Negligibly. The cleaning step is part of the normal pipeline and adds a few seconds at most. You won't notice the difference in total processing time.
-
Will cleaning change what I said?
It removes filler words and hesitation markers — not content. If you said "the API returns three values", that stays. If you said "um, the API, uh, returns, like, three values", you get "the API returns three values".