# AI clip selection: why each clip was chosen Most AI clippers hand back as many clips as they can. Snipri, an [AI clip maker](https://www.snipri.com/) for podcasts and long videos, scores every candidate, drops the weak ones and shows you the scores, so you spend your time posting clips rather than sorting them. - Page: https://www.snipri.com/features/ai-clip-maker - Updated: November 2, 2026 ## What is AI clip selection? AI clip selection is how an AI clip maker decides which moments of a long video become clips. Snipri gives every candidate 5 scores from the transcript and keeps only those whose hook, payoff and standalone sense reach 7 out of 10, then reframes each to 9:16, captions it and says why it was kept. ![Cover for AI clip selection: a vertical clip of a scene from the film Tears of Steel, captioned by Snipri, beside the bar every moment has to clear](https://www.snipri.com/features/ai-clip-maker/cover.webp) - **5 scores**: Hook, Payoff, Standalone sense, Arc and Quotability - **7/10**: The bar hook, payoff and standalone sense must each reach - **Up to 10**: Clips from each job, fewer when fewer moments pass - **16 kinds**: Of video, each judged by its own rules, or Auto - **300 characters**: To name the moments you want it to look for - **15 to 60s**: Clip length by default, or choose 10 to 30s, 30 to 60s, 60 to 90s or 90s to 3 min ## How Snipri picks each clip ### Say what you want, or leave it on Auto Before a job you can pick the kind of video, the clip length and up to 300 characters of moments to look for. On Auto, Snipri names the kind of video from the transcript. ![Drawing of the choices the Studio offers before a job: 16 kinds of video, 5 clip lengths and a field for the moments to look for](https://www.snipri.com/features/ai-clip-maker/studio-options.webp) *The choices before a job, drawn from the Studio's own lists.* ### Every candidate gets five scores Hook, payoff and standalone sense must each reach 7 out of 10. Moments that pass are ranked by hook and payoff, with standalone sense, arc and quotability breaking ties. ![Two moments of one real job scored on hook, payoff, standalone sense, arc and quotability: the clip it kept clears the bar of 7 on all three gates, the moment it left out misses it on hook alone, with a higher payoff than the clip](https://www.snipri.com/features/ai-clip-maker/score-gate.webp) *Two moments of the job on the home page, a clip and a moment it left out, against the bar the worker applies.* ### Only the moments that pass are cut A job returns up to 10 clips, and fewer when fewer moments pass. Each comes with its scores and a written reason, under Why this clip in My clips. ![Timeline of a real job: a 46-minute fireside chat with the 7 clips Snipri kept numbered where they were cut from, ticks at the moments that came closest, and each clip's title and range](https://www.snipri.com/features/ai-clip-maker/one-job.webp) *One real job from the home page: what was kept from a fireside chat, and where.* ## One real job **One of the clips it kept.** Cut from 41:11 of a 46-minute fireside chat, one of the 7 moments Snipri kept from it, with the captions the job gave it. *From “Fireside Chat: Godot's Next Milestones” by Ben Vehling, Matt Ellis and Rémi Verschelde, GodotFest 2025 (media.ccc.de, CC BY 4.0), cut, reframed and captioned by Snipri, with the logo on the speaker's shirt painted out.* ## On podcasts, webinars and interviews - [Podcast clip maker](https://www.snipri.com/tools/podcast-clip-maker): On Podcast it keeps the question when the answer needs it and leaves out intros, sponsor reads and “before we start” housekeeping. - [Webinar clip maker](https://www.snipri.com/tools/webinar-clip-maker): On Webinar it keeps a framework with its example and leaves out agendas, housekeeping, logistics, and recording or slide notices. - [Interview clip maker](https://www.snipri.com/tools/interview-clip-maker): An answer becomes a clip only when it carries enough context to stand alone, so you get whole answers rather than fragments. ## What AI clip selection does not do - It does not predict views. The scores say why a moment works on its own, and there is no virality percentage. - It does not pad a job. A recording with two strong moments gives two clips, and one with none does not count as a video. - It reads what is said. A moment carried by the picture alone, with nothing said about it, is not one it can find. - It does not post for you. You download the clips and post them yourself. ## Frequently asked questions ### What does the AI look at? The transcript, with word-level timings. It looks for complete spoken thoughts with a strong opening, a clear payoff and enough context to make sense alone. You can also tell it the kind of video, how long each clip should run and the moments to look for; a moment you ask for still has to stand on its own to be kept. ### Does it predict how many views a clip will get? No. Each clip comes with its five scores and a written reason for keeping it, which say why the moment works on its own. There is no virality percentage, because nothing in a transcript can promise views. ### Do I need editing experience? No. Clips arrive captioned and framed. The editor is there if you want to change something. ### How do I pay? By the video, not by the clip: a recording counts as one video for each hour of it, rounded up, whether it gives you three clips or ten, and editing, re-rendering or splitting a clip afterwards is free. ### Is Snipri free? You start with 2 free videos of up to 1 hour each, once, with a watermark and no credit card. Paid plans start at $29.90/month for 10 videos a month, each up to 2 hours, without a watermark. A video counts once for each hour of it, rounded up, and only when it gives you at least one clip. ## More Snipri features - [Auto reframe](https://www.snipri.com/features/auto-reframe): Auto reframe turns horizontal video into vertical 9:16 clips. Snipri finds the speaker in each shot, stacks two or three people, or keeps the whole shot. - [Animated captions](https://www.snipri.com/features/animated-captions): Animated AI captions for vertical clips: every word timed to the moment it is said, in a style you pick, in 25 languages, and restyled any time after export. - [Brand kit](https://www.snipri.com/features/brand-kit): Save your logo, handle, font and caption look once in Snipri's brand kit, and every new clip is born wearing them. Change the kit any time, on every plan, free. - [Clip editor](https://www.snipri.com/features/clip-editor): Snipri's clip editor: trim, split or extend a clip, rewrite its hook title, restyle the captions, add text, images, B-roll and music, then re-render it free. ## Guides and comparisons - [How to Make Podcast Clips That Stand Alone](https://www.snipri.com/blog/how-to-make-podcast-clips) - [5 Best Opus Clip Alternatives in 2026](https://www.snipri.com/blog/5-best-opus-clip-alternatives) - [Opus Clip pricing vs Snipri](https://www.snipri.com/alternatives/opus-clip) - [Klap alternative](https://www.snipri.com/alternatives/klap) - [Vizard alternative](https://www.snipri.com/alternatives/vizard)