Your audio, transcribed with speaker labels.
Recent revisions are kept. Restoring one creates a new revision.
