How the clips get chosen
The recording is transcribed first, so the cuts follow what is actually said
Moments that make sense on their own are preferred over arbitrary chunks
Every clip arrives with its timestamp and a line on why it was picked
Download each clip as a file; there is no timeline to export out of
The stages are named while it runs, so a long video never looks like a hang
Your upload is processed to produce the clips and never used to train a model
Use cases
Podcast episodes that owe shorts every weekWebinars and conference talks nobody will watch end to endStreams, where the good two minutes are buried in three hoursInterviews and panels, for a pull quote with the video attachedCourse recordings you want to preview before anybody enrolsLong calls where one answer is worth sending on by itself
Auto Clipper, answered
How long does the video have to be?
At least a minute, because anything shorter has nothing to clip out of it. The page states the file size limit before you pick a file.
How does it decide what to cut?
It reads the transcript and looks for passages that still make sense with nothing around them: a complete point, a question and its answer, a story with an end. Each clip tells you why it was chosen.
What if nothing comes back?
Then it heard no speech, or nothing in the recording stood on its own. Clear, self-contained points work best; music and ambience alone give it nothing to work from.
How much does it cost?
It runs on credits, and the credit cost is shown before you run it. New accounts start with free credits and there is a free allowance every day.