Transcript checker

Paste a transcript or open a .txt, .srt or .vtt file. You get a speaker table and a list of lines to fix, with the time of each one so you can jump to it in the recording.

How it works

Automatic transcripts and fast typing leave the same few faults: a timestamp that goes backwards after a copy-paste, “Dr. Kim” in one place and “Dr Kim” in another, [inaudible] marks nobody went back to, and the speech-to-text loop where one sentence is repeated three or four times.

How to use it: lines can start with a time ([00:01:23], 01:23 or (1:23:05)) and a speaker (Name: text). Both are optional. Set how long a silence between two time marks should be flagged, then press Check transcript. Problems are things a client will notice (time going backwards, repeated sentences, TODO left in). Check rows need a person: a long gap can be a real pause.

The speaker table shows how many turns each person has and their share of the text. A name with only one or two turns next to a similar name is usually a typo. This checks the file, not the audio.

FAQ

Is my transcript uploaded?

No. It is read inside your browser and never sent anywhere.

Which formats work?

Plain text with optional [hh:mm:ss] or mm:ss times and “Name:” speakers. SRT and VTT numbers and --> lines are skipped and the text is checked.

Which languages?

Any. Korean unclear marks such as [청취 불가] and [불명확] are recognised too.