Even a readable transcript may need corrections to names or speaker labels before anyone relies on it. Editing and exporting the result are part of the workflow.
What would you test in a trial? Would you value a quick correction interface over a long list of automated extras?
.
