Remove filler words from your speech
Stop editing out your own ums and ahs. Record, and get back a clean version.
The demo uploads your audio to our servers, where third-party AI service providers transcribe and rewrite it on our behalf.
Preview mode — the backend isn't connected yet, so this demo returns sample text.
That's all 3 rounds of the web trial — in the App, your idea keeps growing: capture anytime, and every take weaves into the ones before it.
Download on the App StoreWhat is this
Um, ah, you know — these aren't content. They're noise your mouth makes while your brain catches up. Every recording is full of them, and cleaning them out by hand is the hidden tax on thinking out loud.
unmuss doesn't just delete the filler words. It reads the sentence underneath them and rebuilds it the way you meant it: tighter, logical, still in your voice. What you get isn't a censored transcript. It's the clean version you would have written if you'd had time.
How to use
-
1. Record up to 60 seconds, or drop in an audio file (mp3, wav, m4a, webm, ogg — up to 2MB).
-
2. unmuss transcribes, strips the filler, and restructures — false starts, repetitions, and filler words gone; the logic rebuilt.
-
3. Copy the clean text — or record another take and watch it fuse with the first one.
Who it's for — and who it's not for
For
Talks you're polishing, podcast monologues, interview self-introductions — anything where you want to sound like the sharpest version of yourself.
Not for
Situations that need a verbatim or legal-grade record. If every "um" is evidence, keep the raw tape.
What you get vs a raw transcript
| Raw transcript | unmuss |
|---|---|
| "So, um, basically what I'm trying to say is, like, our churn went up because, you know, onboarding was, uh, kind of confusing?" | "Churn went up because onboarding was confusing." |
| "I, uh — I think we should, you know, maybe push the launch? Like, two weeks? If, um, design is ready?" | "We should push the launch by two weeks, if design is ready." |
What you just tried vs what the App does
The demo above gave you 3 rounds of single-session cleanup. The App is where the idea keeps growing: Idea Sessions fuse takes across days, version diff shows every change side by side, and you can capture from Apple Watch, Action Button, or hands-free while driving.
Frequently asked questions
- Will it change what I meant?
- No. Filler removal targets noise — ums, false starts, repetitions — not meaning. The logic gets rebuilt, your intent stays. In the App, every pass is saved as a version, so you can always compare or step back.
- What counts as a filler word?
- Um, ah, uh, you know, like, sort of, kind of, basically — plus the half-started sentences and repeated phrases people produce while thinking. unmuss handles all of it in one pass.
- Does it work on files I already recorded?
- Yes. Drop in mp3, wav, m4a, webm, or ogg (up to 2MB). In the App you can also import audio straight from Voice Memos via the Share Extension.
- What languages are supported?
- Eight output languages: English, German, French, Spanish, Portuguese, Japanese, Chinese, and Cantonese. Speak in yours; the AI writes in it.
- How is my voice data handled?
- The web demo uploads your recording to our servers, where third-party AI service providers transcribe and rewrite it on our behalf. You're asked for permission before your first upload. On-device live transcription is an app-only feature. Full policy: /privacy.
Sound like you on your best day.
Download unmuss and stop paying the filler-word tax.
Download on the App Store