Import from Whisper format file

DOTE is able to import transcripts prepared by most Whisper-based AI transcription services, specifically .JSON files in the following formats produced by: OpenAI Whisper, WhisperX, Faster-Whisper, Whisper.cpp., as well as .dote exports from MacWhisper.

Watch the video tutorial on YouTube.

1. Select File ➔ Import or click the Import button on the right of the ribbon bar.
2. Click on the relevant Continue button for "Whisper Format".
3. Specify the "Source File" location for the Whisper file to import.
4. Enter a unique name for your new Transcript Name.
5. Choose from the Whisper Import Options:
   • Transcript Convention - Jeffersonian or Mondadaian.
   • Default Speaker Designation - In case there is no speaker-id supplied by Whisper, the default speaker designation will be added on each line.
   • Create Sync-codes from Timing Information.
     • Start of line - The timing information is used to create a sync-code representing the start of the line.
     • Start -> End range - Use the duration from the timing information to generate a ranged sync-code.
     • Include timestamps on lines with sync-code
       • Timestamp format - Specify the precision of the timestamp
       • Timestamp placement - Specify where the timestamp should be located on each line
       • Timestamp brackets - Define which type of brackets should be used around a timestamp
     • Add silence markers for long pauses - For pauses longer than 1 second, add a line showing the pause. Example: (1.5)
   • Include Timestamps for each line - Add a technical comment to each line showing the timing information provided for that line. Example: // [Time: 0:22 - 00:27]
   • If per-word translation confidence values are provided:
     • Print average confidence % per line - Display this information as a technical comment at the end of each line. Example: // [Avg. Confidence: 85%]
     • Minimum confidence threshold - Set a percentage value where if the translation confidence is lower, wrap the word or group of words in parentheses. Example: hello my name is (jane)
     • No confidence threshold - Set a percentage value where if the translation confidence is lower, replace the word or words with a spaces (one space per 0.1 seconds of duration), wrapped in parentheses. Example: hello my name is ( )
   • Maximum Line Width - If a line exceeds this, then it is wrapped to a new line without an additional speaker designation.
   • Remove Punctuation - Remove any punctuation symbols (".,?"). NOTE: this will also remove those symbols even if they are used to mark interactional intonation (though whisper is currently unable to produce such interactional markers)
   • Remove Sentence Capitalisation - Remove from the body of the transcript to be imported. NOTE: this will also decapitalise proper nouns and abbreviations if they appear at the start of a sentence.
   • Width of Name Column - Number of characters between start of line and body of transcript.
6. When ready, click on Start Import.
7. The Transcript will be added to the current Project, so make sure the audio/video file used to prepare the whisper export is comparable to the transcript.

A preview of the imported transcript will be updated on the right side after any changes to the Options.