We use automatic recorded lectures everywhere, and you have an editor functionality. I propose you could combine some of the automatic transcription, image recognition, and sound analysis to make a best-guess determination of when a lecture starts and ends. Give us a per-course setting to "automatically try to trim beginning and end", which would do this detection and edit.

Our faculty start within a couple minutes of the recording and try to end right at the bell (and we usually have the recording go over by a few minutes as a buffer) leading to confusing starts of videos, and endings that include snippets of private conversations as students file out of the room.

When a lecture starts, there's a characteristic change to sound in the room (students go quiet, one clear speaker begins), the screen feed changes its image to a full-screen ppt or document camera, on the camera feed an instructor's face suddenly looks up and out to the audience, and the words sound like the start of a lecture.

When it ends, presentation mode is closed, maybe the PC logs out or content on a document camera is removed, the general din of noise goes up, and there's just been some concluding remarks.

(And, as a hard limit, these moments occur near the start and near the end of the recording; group activities in the middle of class shouldn't be marked as an ending).

I think with all these indicators, a practical time-saving feature could be made. Since edits in Echo360 are non-destructive, it should be possible to re-process without the edits if something gets cut that shouldn't.