File formats and audio
Timestamp
A time reference tying a piece of the text to a particular point in the audio recording.
A timestamp ties a piece of the transcript to a particular point in the audio, typically written as 00:14:32.
They look like a detail, and are not.
What they make possible
Finding things again. Without timestamps an 8,000-word transcript is just text. With them you can jump straight to the place in the recording and hear what was actually said.
Checking quotes. For journalism, research and legal work this is the difference between a quote that can be defended and one that cannot.
Captioning. SRT and VTT are in practice text with timestamps. Without them captioning is impossible.
Editing efficiently. If you see something odd in the text, it takes two seconds to hear the place instead of scrubbing for it.
Resolution
- Per segment or sentence. The most common, and enough for most purposes.
- Per speaker. At every change of speaker. Enough for meeting summaries.
- Per word. Necessary for captions that follow speech closely, and for searching audio precisely.
Keep them
People often paste the transcript into a document and lose the timestamps along the way. Keep a version that has them, whatever you do with the text afterwards. Recreating them later means running the whole recording again.
See also
SRT, VTT, captioning, diarisation.