In the tool, when clicking the "new transcription" button, you will need to enter some data:
- Project description
- presenter (optional)
- Audio file (supported are .wav and .ogg)
- Transcript file (supported are .ctm and .srt)
- .srt: segment-aligned subtitle format (see here for details).
- .ctm: word-aligned transcription format. Each line corresponds to one word and contains space-separated fields: transcription identifier, #channels, start-time (sec), duration (sec.), word, and optional confidence score. The first two fields are not used by our tool. We treat comment lines (starting with #) as indicating a new segment. If further the comment line contains the string "__INACTIVE_SEGMENT__", this segment is displayed in gray and should not be modified. This can be used for example to transcribe/correct only the segments that are likely to contain errors. An example can be found here. Note that filler words such as <breath> are ignored, and that pronunciation tags such as "the(1)" are removed: "the".
To transcribe from scratch, select the checkbox. In this case, no initial transcript is needed, and it is sufficient if the .ctm file contains segments (lines starting with #) and filler-words to indicate time and duration of the segment (e.g.: "x 1 10.0 5.0 <empty>" for a segment starting after 10 seconds and lasting 5 seconds).