Filed under: Accessible communication
A workshop slide lists two possible session times. The chair says, “Let’s examine the second option.” A reader of the recognised speech gets the words correctly, yet still cannot tell which time is being discussed.
This is a fictional example of a real editorial distinction. The W3C WAI guidance on transcripts describes a basic transcript as carrying speech and meaningful non-speech audio. A descriptive transcript also carries visual information needed to understand the content. Correcting the speech-recognition output cannot recover information that was never spoken.
Add the missing channel without inventing speech
Suppose the slide in our example lists 09:00 and 13:00. Compare these two records:
Speech alone: Chair: “Let’s examine the second option.”
With the necessary visual information:
Visual note: The slide lists session options 09:00 and 13:00.
Chair: “Let’s examine the second option.”
The second version lets a reader connect the spoken reference to 13:00. It does not put those numerals into the chair’s mouth or turn a discussion into agreement on a time. The added information is labelled, so the reader can distinguish the source channels.
If the real slide were unreadable, a plausible guess would not solve the problem. The editor would need a reliable, authorised copy or a clear note that the reference remains unresolved. The same principle applies to an unidentified speaker: clarity should not come from supplying a name the evidence cannot establish.
Turn timed fragments into a readable passage
Caption lines are arranged for synchronised reading alongside a recording. A transcript can instead group a speaker’s related remarks into paragraphs. Keep a separate caption track where it is needed; changing its layout into prose does not make the two formats interchangeable.
For the workshop, a heading such as “Discussion of session times” would help readers find the passage. A useful timestamp could take them back to that point in the recording. WAI presents headings, links and timestamps as possible aids, not an instruction to timestamp every line.
The heading must describe the record accurately. “Session time agreed” would introduce a decision that our passage does not contain. If the organisation needs formal minutes as well, that is a separately identified record with its own review.
Check the reference, then the route to it
Review the final transcript against the recording version it accompanies. Follow references such as “this,” “over here” and “the second option” to see whether the text gives the reader what they need. Retain meaningful sounds and visual information; mark unresolved points.
Then make the transcript easy to find from the recording page. A carefully prepared account has little practical value if the person looking for it cannot locate it.
For this example, success is specific: the reader can identify the option under discussion and can see that no decision has yet been recorded. That is a better result than a flawless copy of words that still leaves the answer on the slide.
The workshop, times and dialogue are invented. The example assumes a recording the organisation is authorised to retain and share; a transcript does not create new distribution permission.