AutocensorAUDIO · API
DemoFeaturesStoryStartSign inUpload audio

Web platform · podcast & production audio

Clean audio.
Cut the bad words.

Upload a file. Get word-level timestamps, automatic profanity detection, and a cleaned track — beep it, or remove it — without opening a DAW.

Upload audio →Create accountfree to start · no account required for a first run
live demo · waveform + transcriptRAW · uncensored
Toggle how Autocensor handles a flagged segment on the track.
00:14

So the whole pitch was fine until someone said f**k and the room went quiet.

clean speech flagged word beep fill removed / stitched

// capabilities

Everything between upload and a clean master.

Transcription, alignment, and audio surgery in one queue — built for podcast pipelines, not generic “AI audio” toys.

ASR

Word-level transcription

Forced alignment with confidence scores so every beep and cut lands on real boundaries — not guessed silence.

path · audio → ASR → timestamps → EDL
CENSOR

Profanity detection

Flag target language in the transcript, then apply the edit on the waveform with crossfades so you don’t hear a hard cut.

modes · beep · remove · keep-raw
BEEP

Broadcast-style beeps

Classic 1 kHz fill over the exact word duration. Useful when you need the airtime preserved for pacing.

tone · 1 kHz · duration-matched
EDIT

Silence & dead air

Strip awkward pauses and tighten speech cadence while keeping natural rhythm — not a nuclear “remove all silence” pass.

goal · tighter take · same voice
LIVE

Realtime job status

WebSocket updates from queue → processing → complete. No refresh-spam while a long episode cooks.

transport · WS · job events
GUEST

No account required

First runs don’t need a signup. Create an account when you want history, podcasts, and billing.

start · /upload · free trial path

// the cut

Manual scrubbing vs. an edit decision list.

The old way burns an hour of producer time per episode. Autocensor turns the same problem into a job you can queue.

OLD WAYtimeline scrub

You listen at 1.5×, mark the swear, razor it, drop a beep, check the crossfade, export, re-listen for the pop you just introduced.

1. scrub → find word
2. razor · fade · beep sample
3. bounce · QC · hope
cost: your afternoon
AUTOCENSORtimestamped EDL

Transcribe once. Flag once. Render a cleaned master with the mode you chose — beep or surgical remove — while you move on.

  • 01Word timestamps drive the cut points
  • 02Beep or remove without reopening a DAW
  • 03Download cleaned audio + JSON transcript

// requirements

What you need to ship a clean file.

Input
MP3 · WAV · M4A · FLAC
Upload from the web UI or wire the API later.
Browser
Any modern desktop
No plugin. Status over WebSocket when processing.
Account
Optional at first
Guest upload works; accounts unlock history & podcasts.
Price
Free to start
Pay when you need volume — not for a single test file.

// get started

Three steps. No install.

  1. Drop a file

    Open /upload, choose options (beep, remove, silence), and send the job.

  2. Watch the pipeline

    Transcribe → align → edit. Live status on the job page while workers process.

  3. Download the clean master

    Grab processed audio plus the timestamped transcript. Account optional for history.

READY · WEB

Try Autocensor on a real file

No app download. No credit card gate for a first pass. Upload, censor, download.

guest okapi availablews live status