By ShinobiTools Team · Last updated: July 2026
Upload a recording and get a clean version back: every swear word beeped, muted or cut out on the exact word. You see the full list with timestamps first and decide word by word. Free, no account, no watermark.
Transcribe audio now →The censoring itself is easy. Finding the words is the job: scrubbing an hour of audio, dropping a marker on every f***, lining a tone up with a word that lasts a third of a second, and doing it again when you miss one. An hour of podcast is an evening in an editor.
ScribeGrab flips the order round. It transcribes your file first, and Whisper hands back a start and end time for every individual word. So the swearing is already found and already timed before you touch anything: all that's left is your decision about what happens to it.
Three options, and the difference matters more than it sounds:
Every option lands on the word itself, plus a hair either side, because a transcriber's word boundary can sit a few hundredths of a second too tight.
Automatic censoring goes wrong in the same way every time: it doesn't know a name from an insult, or a quote from an outburst. So ScribeGrab shows you the list before it changes anything — every hit with its timestamp, written as f*** — and one click leaves any of them in.
Milder words (d***, h***, c***) sit behind their own switch and are off by default. A swear in a podcast is not the same problem as a swear in a school video, and that's your call, not ours.
If you only beep or mute, nothing happens to the picture at all: the video stream is copied over untouched, so there is no second generation of compression and no quality loss. Only cutting forces the picture to be re-encoded, because the frames themselves change.
Audio comes back as MP3; video comes back in its own container (MP4, MOV, MKV, WebM), or as MP4 if you chose to cut.
The strong list covers the everyday English swearing and slurs plus the most common Dutch, German, Spanish and French words, matched on whole words only — so "assessment", "hello" and "cocktail" are never touched. It is deliberately not a complete dictionary of everything anyone has ever said: check the list it shows you before you download.
While you're there you can also cut the "um"s and the long silences out of the same file, and the transcript, subtitles and speaker labels come with it either way.
Yes. Up to 90 minutes and 2 GB per file, no daily cap, no account and no watermark on the result. Ads pay for the GPU.
Yes. The list of found words is shown with timestamps before anything is rendered; click a word and it stays in the recording exactly as it was.
Not if you beep or mute — the picture is copied over bit for bit and only the audio is rebuilt. Cutting does re-encode, because removing frames changes the video itself.
It only knows the words on its list, and a very muffled word can be transcribed wrong. You always get the transcript too, so search it if you want to be sure.
No. Ticking the clean-version box keeps your file on the server for this job instead of deleting it straight after transcribing — and then everything, original and result, is wiped within 45 minutes. See privacy.
Whisper transcribes about 100 languages. The swear-word list itself is English plus the most common Dutch, German, Spanish and French words.