Turn a Whisper caption export into a usable editing file
Whisper output can be a starting point for a caption workflow. Bring the resulting SRT or VTT to the assistant to discuss conversion, shifts and delivery formats.
Try this in the assistantThis guide is in English. The assistant supports 83 interface languages.
Separate text mistakes from timing mistakes
Read the transcript against the recording before building a final export. A wrong word needs a text correction; a caption that starts late needs a timing correction. A converter can change the file structure without knowing which words were actually spoken or sung.
Explain how the recording changed
A cut at the beginning can cause a consistent offset. A speed change can cause progressive drift. Several removed sections need local edits rather than one global shift. Tell the assistant about those edits so it can suggest an appropriate workflow.
This Whisper SRT is for the original video. My edit removed the first 2.4 seconds. Help me shift it and export VTT.
Supplied lyrics are a different starting point
Transcription estimates the words in a recording. Aligning exact lyrics starts with text you already know. For music where every sung word matters, the dedicated LyricTimestamps workspace lets you align supplied lyrics and refine the result.
Questions about this workflow
Does this page run Whisper?
This guide covers working with an existing caption export. It does not promise a hosted Whisper transcription endpoint.
Can conversion fix incorrectly transcribed lyrics?
No. Correct the text using the recording or your verified lyric sheet. Repackaging the file does not verify the words.
Can I keep the original file?
Yes. Keep the original export and save repaired versions separately so you can compare timing and wording.