Skip to content

Video highlight finder

Find the liveliest moments in a stream VOD, a podcast or any long video by its sound: the laughs, the shouting, the hype. Jump to each highlight, play it, and save it as a clip. It runs on your device, so nothing is uploaded.

A free tool from Clyper, the AI clip maker that turns long videos into captioned vertical clips.

Get started

Drop a stream VOD, a podcast or any long video here

It listens for the laughs, the shouting and the hype, and marks each one.

Clips

Clip length

Each clip starts a few seconds before it gets loud, in a pause when there's one near, so it doesn't open mid-word. None of them overlap.

It hears loudness, not meaning. Laughter, shouting, a cheering crowd and sudden hype stand out; a great line said quietly doesn't, and a loud song or ad counts like a laugh. Play each one before you post it.

MP4, MOV, WebM or MKV, or MP3, M4A and WAV. It's read on your device: nothing is uploaded.

Free, no watermark, nothing to sign up for. Files up to 6 hours long: a whole stream VOD.

How to find the highlights in a video

  1. Choose the recording, or drop it on the player: a stream VOD, a podcast episode, a long video, or just its sound. Files up to 6 hours work.
  2. Wait while it listens. Only the sound is read, so it takes far less time than the recording runs. The highlights appear on the bar under the player and in a list, best first.
  3. Pick a clip length: 30, 45, 60 or 90 seconds (45 to start). The clips follow it straight away.
  4. Play each highlight, from the list or the bar. Keep the ones that make sense on their own.
  5. Save clip downloads that part of the file, or press Copy timestamps to take the whole list to your editor.

How it finds the best moments

It listens to the loudness, not the words. It measures how loud the sound is 50 times a second, then compares every moment with the two minutes around it. That's why a quiet podcast and a loud stream both get highlights: what counts is how much louder a moment is than the talk around it, not how loud it is on its own. If the mic was turned up halfway through, the moments in the quieter half still count.

A moment ranks higher the further it rises, the longer it holds (a single bang counts for less than five seconds of laughter) and the quieter it was just before, so a sudden burst after a calm stretch stands out. Silence never rises, and neither does steady sound like music under a stream, so they're left out.

Each highlight becomes a clip of the length you pick, starting a few seconds before it gets loud (in a pause when there's one near, so it doesn't open mid-word). Clips never overlap. You see about one for every three minutes of the recording (up to 20 at first), and can ask for more.

What the reasons in the list mean
The list saysWhat it heardOften
A burst after a quiet stretchMuch louder, right after the seconds before it were quietA jump scare, a sudden shout, a big play
Much louder than the two minutes around it9 dB or more over the talk around itA big laugh, yelling, a cheering crowd
Stays loud for 20 secondsLoud for 12 seconds or longerA long laugh, an argument, hype that keeps going
Louder than the two minutes around it5 to 9 dB overLaughter, raised voices, excitement
A little louder than the two minutes around it3 to 5 dB overEmphasis: worth a listen, not always a moment

What it can't hear: a great line said calmly, a look, a clever play in silence. And a loud song, an ad, a donation alert or an explosion in a game counts the same as a laugh. Treat the list as where to look first, not as the clips to post.

Picking the clips worth posting

  • Listen to what comes before the noise. The laugh is the payoff; the story or the joke before it is the clip. If the setup starts earlier than the clip does, choose a longer length or move the start with the video trimmer.
  • Keep the ones that stand on their own. Someone scrolling hasn't seen the stream. If a moment needs five minutes of context, skip it, however loud it is.
  • Start on the first words. The opening second decides whether people stay, so trim any dead air at the start. More in how to make clips that go viral.
  • On a stream, keep the reaction in the frame. The webcam is what makes a gaming moment work vertically: put it over the game with the split screen video maker.
  • On a podcast, follow the laughter back to the story. A run of highlights close together is often one good conversation: listen to the whole stretch and cut the best minute of it.

Then make each clip 9:16 with the vertical video converter and caption it with the auto caption generator. For a podcast that's only sound, the audiogram maker turns a clip into a video.

Getting the whole stream or episode

From Twitch, download a past broadcast from the Video Producer, under Content in your Creator Dashboard. Twitch keeps past broadcasts for 7 days for most streamers, 14 days for Affiliates and 60 days for most Partners and Prime and Turbo users, and only if Store past broadcasts was turned on before you went live, so download the streams you want to clip before they go.

From YouTube, download your own video in YouTube Studio: Content, then the video's menu, then Download. It comes as an MP4 at 720p or 360p, up to five times a day for each video, so if you have the original recording (from OBS, Riverside or your camera), use that instead.

Clipping someone else's stream or show? You need their permission, which clipping campaigns give. See how Clyper handles stream clips, podcast clips and Twitch VODs.

Questions

How do I find the best moments in a video?

Choose the video here and the highlight finder listens to its sound, then marks the moments that get much louder and livelier than the minutes around them: laughter, shouting, a crowd, hype. Play each one, keep the ones that work on their own, and save them as clips.

How do I find highlights in a Twitch VOD?

Download the past broadcast from the Video Producer in your Creator Dashboard, then open the file here. It works on up to 6 hours at a time, and only the sound is read, so it takes far less time than the stream ran. Twitch deletes past broadcasts after 7 to 60 days, depending on your account, so download the ones you want to clip.

Does it work on podcasts?

Yes, on the episode's video or on the sound alone (MP3, M4A or WAV). Because each moment is compared with the minutes around it, a quietly recorded podcast has highlights just like a loud stream: the big laughs and the raised voices.

How does the highlight finder decide what's a highlight?

By loudness over time. A moment ranks higher the further it rises above the median of the two minutes around it, the longer it holds there, and the quieter it was just before. Silence and steady sound, like music under a stream, never rise above that, so they don't count.

What can't it find?

Anything that isn't loud. A great line said calmly, a funny look or a clever play in silence won't stand out, and a loud song, an ad or an alert sound counts like a laugh. Clyper reads what's said as well, so it finds those quieter moments too.

Is my video uploaded anywhere?

No. Your browser reads the sound, finds the highlights and saves the clips, all on your device. Nothing is sent to us or anyone else.

Is it free, and do saved clips lose quality?

It's free, with no watermark and nothing to sign up for. A clip from a video is copied out of the file as it is, without re-encoding, so it's the original's quality and saves in seconds. From a sound file, the clip is saved as an MP3.

Sources

Read on October 26, 2026.

Find the moments in the words, too

The highlight finder hears where it gets loud. Clyper also reads what's said, so it finds the moments worth posting, loud or quiet, and does the cutting, framing and captions for a whole video: up to 30 clips from one link or upload.

or upload a video

See plans and pricing

Guides

More free tools