VoxScrub strips background noise, filler words, dead air and mouth sounds from your recording automatically — so you spend your time podcasting, not scrubbing a timeline.
No sign-up needed to test it below.
Before
After VoxScrub
Illustrative example — try it yourself with the live demo below
Upload once. VoxScrub listens for the six things that make an episode sound unedited, and fixes them together.
Honking cars, barking dogs, hum and hiss — gone, without re-recording a single line.
So um, today uh we'll like dive right in.
"Um", "uh", "like" and repeated stutters, cut cleanly across 20+ languages.
Silences over your set threshold shrink automatically, keeping natural breathing room.
Clicks, smacks and heavy breaths lowered or removed without flattening your voice.
Auto-generated show notes and chapter markers, ready to paste into your description.
Clean every guest's track separately, then merge into one uninterrupted episode.
This runs for real, in your browser — no upload to a server. Drop in a voice recording and VoxScrub's noise-reduction engine lowers the background noise while keeping the voice's timing exactly as it was — nothing gets cut or shortened.
Choose a template — same as Cleanvoice's pre-made presets, each runs a real, different pass
Log in to try the live demo
Free, takes a few seconds — your file still never leaves your device.
This runs a real spectral noise-reduction pass live, using your browser's audio engine — pick a template above and it actually changes the processing strength, brightness boost and whether loud/broadband (music-like) passages get touched. After it finishes, you'll land on a results page with a before/after and a strength slider. "Keep music same" uses a simplified loudness-based heuristic, not true music/voice separation. Filler-word detection and full transcription are on the roadmap and not live yet.
Audio or video, any length. Batch upload if you've got a whole season to clean.
Noise, fillers, dead air and mouth sounds are detected and fixed in one pass.
Download the clean file, or export a timeline of edits into your usual editor.
Editing, enhancement, transcription and summarization — through a single endpoint. No hours-based limits, no file-size ceiling.
Get notified when it shipsVoxScrub started from a simple annoyance: recording an episode takes an hour, and editing it used to take three. Every "um," every siren outside the window, every ten-second silence while someone found their notes — all of it had to be found by hand.
So we're building the editor we wished existed: one that listens to a raw recording the way a producer would, and fixes what a producer would fix, automatically.
Try what we've built so farEvery feature starts from "what would waste a podcaster's afternoon," not "what's technically impressive."
Files stay yours — processed on request, removed after a short retention window, never used to train on without asking.
Every automatic cut can be reviewed, undone, or exported as a timeline for manual fine-tuning.
Yes — the noise reducer above genuinely analyzes the audio file you upload, learns what your background noise sounds like, and reduces it across the whole clip, right in your browser. The voice's length and timing stay untouched. It's a working slice of the full product, not a mockup.
Background noise, filler words, long silences, mouth sounds and breaths — matching Cleanvoice's feature set, with our own detection pipeline and interface.
Planned: pick a preset template, or toggle individual features on and off — for example, trim silence but leave filler words untouched.
If you sign in with email, a real account is created on this site's own server: your email and a securely hashed password (bcrypt — never plain text) are saved in a database. "Continue with Google" isn't connected to real Google sign-in yet. Your audio itself never touches any server either way — it's processed entirely in your browser.
Yes, a single-endpoint API for edit, enhance, transcribe and summarize is on the roadmap, shown in the developer section above.
No sign-up. No file leaves your device. Just upload and listen.
Jump to demo