Highlight ClipperBETA

Auto-find the highlights in any video — by ear

Highlights have a sound. A goal, a dunk, a clutch round, a big reveal on stream — they're almost always wrapped in a spike of energy: the crowd erupts, the commentator's pitch jumps, the room reacts. The Highlight Clipper turns that into a shortcut. It extracts the audio from your video right in the browser, measures the loudness moment-to-moment, and flags the sustained spikes that stand out from the baseline. Instead of scrubbing a two-hour recording for the three moments worth posting, you get a ranked shortlist and jump straight to each one.

What it's great for

How the detection works

The tool decodes your video's audio and walks it in short windows, computing the loudness of each in decibels. It establishes a baseline (the typical loudness of the recording) and then looks for stretches that rise above it by your chosen sensitivity and stay elevated for at least the min sustain time — the signature of a crowd roar rather than a one-off clap. Each surviving moment is scored by how loud and how long it spiked, ranked, and turned into a proposed clip with your lead-in and clip length. No machine-learning model, no cloud, no guesswork you can't see — you watch the energy curve and the picks on the same timeline.

Notes & limits

• It listens, it doesn't watch. A quiet goal with no reaction won't spike; a loud non-event (a whistle right by the mic) might. The sliders let you tune it, and you confirm each pick in the player.
• Big files use memory. Analysis decodes the audio into memory, so a multi-hour 4K file can be heavy on a phone. A clip of a half or a single period is the sweet spot for now.
• Everything stays local. Your video is never uploaded.