On-device transcription finds spoken PII — numbers, emails, names — and you mute the segment containing each one (the review shows every span and the total). The audio never leaves your browser, and you get a masked transcript too.
On-device transcription finds spoken PII — numbers, emails, names — and you mute the segment containing each one (the review shows every span and the total). The audio never leaves your browser, and you get a masked transcript too.
Redact spoken sensitive information in an audio file, entirely in your browser. GhostAudio transcribes the recording on your device with Whisper, detects PII in the transcript — phone numbers, emails, credit-card and IBAN numbers, and optionally names — and shows every finding with its timestamp so you can review it. Uncheck anything that should stay audible, then apply: the transcript segment containing each finding — usually the whole sentence around it — is silenced with a real audio filter (the review shows each muted span and the total muted time), and you can download a transcript with the redacted words masked. The audio never leaves your device, nothing is uploaded, and there is no watermark or signup. Free, private, and dead simple: drop the file, review the findings, mute.
Drop your audio
Drag an MP3, WAV, M4A, FLAC, or OGG file onto the page. It opens in the GhostAudio studio with Mute private speech pre-selected — nothing is sent anywhere.
Scan on your device
Whisper runs in your browser to transcribe the recording, then the same detectors GhostRedact uses find sensitive data in the transcript — numbers, emails, cards, IBANs, and optionally spoken names.
Review and mute
Every finding is listed with its timestamp — uncheck anything that should stay audible, then apply. The segment containing each finding is silenced (the review shows each span and the total), the rest of the audio is untouched, and you can download a masked transcript.
Is the audio really muted, not just hidden?
Yes. The mute is written into the output's audio with a real filter — the silenced segments contain silence, there is no way to 'un-mute' them later. This is the audio equivalent of GhostRedact's rasterize-and-destroy approach for PDFs.
How accurate is the detection?
Transcription is done by Whisper running locally, and detection covers structured data (phone numbers, emails, credit cards, IBANs, IPs) deterministically. Spoken personal names are found by an on-device model (~80 MB) that loads automatically on desktop-class devices; on phones, low-memory devices, or slow and data-saver connections it's offered as an opt-in download instead. Muting works per transcript segment, so a finding mutes the sentence or phrase around it, not just the exact words. Automatic detection is a review aid — listen through the selected moments before applying.
What about privacy?
The entire pipeline — transcription, detection, and muting — runs in your browser. No audio, transcript, or finding ever leaves your device; you can watch the Network tab while it runs.
Which formats can I redact?
Anything the studio opens: MP3, WAV, M4A, AAC, FLAC, OGG, and more. The muted output is delivered as a universal MP3.
Muting spoken details, covering faces and on-screen text, and stripping media metadata.
Every GhostAudio tool runs entirely in your browser. Your file is never uploaded — there's no upload endpoint to send it to.