Take the people
out of a transcript
Drop in a WebVTT, SRT, plain text or Teams transcript. Everyone in it is listed, each one gets a single pseudonym used everywhere they speak and everywhere they are named, and email addresses, phone numbers and links come out with them. Then the tool re-reads its own output and tells you what it could not reach.
Your file never leaves this browser
Reading, matching, replacing and exporting all happen on your own machine. Nothing is uploaded, nothing is stored, there is no account to create, and closing the tab throws the whole thing away.
Drop your transcript here
or choose a file, or paste the text below
WebVTT, SRT, TXT or MD, up to 10 MB
Nothing loaded yet
Drop a file or paste a transcript above. Everyone in it shows up here with a pseudonym already filled in.
Three steps, about a minute
Load the file you already have
A .vtt, .srt, .txt or .md file, or the transcript pasted straight out of a Teams .docx export. It is read in the browser, not sent anywhere.
Approve what comes out
Everyone in the transcript arrives with a pseudonym already suggested. Email addresses, phone numbers and links are listed with their matches, so you see them before they go.
Copy it or download it
Readable turns, Markdown, plain text, or a re-emitted WebVTT file with the original timings. The result is re-read first and anything still standing is reported to you.
What comes out, and how
Everyone the file named
The speaker labels are the participant list, so there is nothing to guess at. Each person becomes one pseudonym, both where they speak and where a colleague says their name mid sentence.
First names, surnames and the full name
A label like "Jordan Blake (Guest)" also covers "Jordan" and "Blake" on their own, which is how people are actually addressed in a meeting.
Email addresses, phone numbers and links
Matched by shape. Phone numbers are caught in international and grouped local formats, and a leftover timestamp is never mistaken for one.
Whatever else you type in
A client, an employer, a city, a codename. One per line, matched as whole words in scripts that use spaces and as plain text in Japanese, where words do not have them.
What it cannot do
This is exact matching, not judgement. It will not catch a name spelled a second way, a nickname nobody wrote down, or a name that grew an ending, although it does tell you when it finds one of those left over. It also cannot know that a role plus a team plus a date still points at one person. Read the result before you send it, and treat this as the first pass rather than the last one.
Before you paste anything
- Does my transcript get uploaded?
- No. The file is read, matched, rewritten and exported by JavaScript running in your own browser. There is no upload, no server call, no storage and no account. Open your network tab and watch nothing happen.
- Is the result anonymous?
- That is not a promise this tool can make on your behalf, and be careful with any tool that does. It removes the names, addresses, numbers and terms you point it at. Whether what is left can still be traced back to a person depends on the content: a job title, a team size and a city are often enough on their own. Read the output and decide.
- Does it use AI to find the names?
- No, and that is deliberate. It reads the speaker labels your file already carries, so the participants are known rather than guessed, and every mention of the same person becomes the same pseudonym every time. Something that guesses is something that is occasionally inconsistent, and inconsistency is the one thing a redaction cannot afford.
- Can I keep a record of who is who?
- Yes, but not here. Type your own replacements instead of the suggested ones and keep that mapping in a separate file, stored somewhere else. This page holds nothing once you close the tab, on purpose.
- Which files work?
- WebVTT (.vtt), SubRip (.srt), plain text (.txt and .md), and a Teams transcript pasted out of the .docx export. Anything carrying "Speaker 1:" style labels or cue timings will parse.
- Can I get a WebVTT file back out?
- Yes, as long as your source had both start and end times on its cues. The WebVTT tab re-emits every cue with the original timings and the redacted text, and you can download it as a .vtt file.
- Is this the only tool that does this?
- No. Other redaction tools exist, including good ones that run in the browser. This one is free with no account, knows who was in the meeting because it read the speaker labels, checks its own output afterwards, and is the same tool in seven languages.
Reline does not redact this for you
Worth saying plainly, since you are standing on our page: redaction is not a Reline feature and this tool is not a trial of one. What Reline does is sit upstream of the job. It records the meeting on your own machine without sending a bot into the call, separates the speakers as it transcribes, and keeps the note to yourself until you decide who else gets it.
The free plan covers 1 hour per note, 10 hours of transcription a month and 2 GB of storage. No card needed.
- Bot free recording: mic and system audio captured locally on your machine
- Speaker separated transcripts in 60+ languages, with the audio kept so you can listen back
- Notes start private. Holding a seat in the workspace does not give anyone access to your note.