Glyphcast
Use casesWebinars

Webinar highlights, without rewatching the webinar

A webinar recording is an hour of which about four minutes is worth sending to anyone. The four minutes are real and they are somewhere in there; the problem is purely that finding them costs an hour.

The Q&A is usually the best material

The presentation is scripted and the audience has heard the pitch. The questions are not scripted, and the answers to them are where a speaker says the specific, concrete thing that the slides had been carefully generalising around.

That part also scores well, and for structural reasons rather than luck: a question and its answer are two voices with a pause between them, which reads as a clean boundary, and the answer tends to arrive with more energy than the section that preceded it.

Screen shares survive the reframe

Most webinar recordings are mostly slides or a shared screen, which is exactly the material a centre-crop to vertical destroys - the text is spread across a wide frame, and a vertical slice of it is unreadable.

Because the 9:16 output keeps the whole frame rather than cropping into it, a shared screen stays legible. It is smaller, because a wide frame inside a vertical canvas has to be, so a dense spreadsheet will not read on a phone. A slide with a headline on it will.

One recording, clips for several channels

The useful output of a webinar is rarely one clip. It is a handful: one for the sales team to send, one for the product page, two for social. Glyphcast proposes a set rather than a single best moment, and the number is capped by your plan rather than by the tool deciding on your behalf that one is enough.

Where a video genuinely only contains two good moments, you get two. The cap is a ceiling, not a target - padding the set out with weaker cuts would just move the reviewing work back onto you.

How it works, step by step

  1. 01Upload the recordingUpload the recording exported from Zoom, Teams, Meet or whatever hosted it.
  2. 02Speech and energy are scoredThe recording is transcribed with word-level timings, and moments are scored on what was said alongside audio energy and on-screen motion.
  3. 03Review and trimClips arrive captioned and reframed. Trim any edge and it is re-cut through the same pipeline, captions and all.
  4. 04Download the setTake them one at a time or the whole set as a ZIP.

Questions

Can it tell the Q&A from the presentation?

Not as a labelled section - it does not segment a recording into named parts. What it does is score every moment on the same signals, and question-and-answer exchanges tend to score well on them, so they usually surface without being sought.

What about a recording that is only slides and a voice?

It works, and the scoring leans almost entirely on the transcript and the audio, since a slide deck offers very little motion signal. The clips will be legible as long as the slides are not dense.

Can I get a clip of one specific answer?

Indirectly. Glyphcast proposes the moments it scores highest rather than taking a timestamp, but any clip can be trimmed, and trimming re-cuts through the whole pipeline so the result is a properly captioned clip rather than a crop of one.