How to caption a lecture for YouTube, Panopto or Moodle
Automatic captions were trained on everyday speech. They write “x squared over two sigma squared” as words, turn Euler into “oiler” and Fourier into “four ier”, and have no idea that “h two o” is a formula. Accessibility offices report 20 to 40 minutes of hand-editing per hour of STEM lecture. Uploading your own subtitle file removes almost all of it.
1. Export the audio
Drop the recording straight in (mp4, m4a, mp3, wav), or export audio from Panopto, Zoom or your lecture-capture system. Anything under 45 minutes runs in the browser; split a two-hour lecture at the break.
2. Pick a notation once
- Unicode (x²/2σ², ∫₀^π sin x dx, H₂O) renders in every caption player and is what most courses want.
- LaTeX ($\frac{x^{2}}{2\sigma^{2}}$) for transcripts that go into lecture notes, Overleaf or a MathJax-enabled LMS.
- Plain (x^2/2sigma^2, sqrt(x)) for screen readers and systems that strip non-ASCII characters.
- Units and operators can each be switched off if your accessibility guidelines ask for spoken forms.
3. Transcribe and review the corrections panel
Every change is listed with a timestamp; click one to hear the moment. Stemscribe typesets what the lecturer said and does not check the maths, so this is where a teaching assistant verifies a slip of the tongue.
4. Download .srt and upload
In YouTube Studio open the video, choose Subtitles, then Add under your language and Upload file with timing. In Panopto use Captions → Upload; Moodle, Canvas and Kaltura accept the same .srt or .vtt. Caption text is indexed for search, so “L’Hôpital’s rule” spelled right is how students find the lecture.