Clip a Twitch VOD to TikTok
Paste the VOD link. Momevera transcribes the whole thing, scores every window against an absolute quality bar, cuts on sentence boundaries, reframes to 9:16 around whoever is talking, and burns word-by-word captions. No download step, no timeline.
Free tier: 90 credits every month, no card. Every source is 1 credit per minute — a 40-minute window is 40 credits.
What happens after you paste the link
The VOD is fetched and transcribed end to end
Not sampled. The whole source gets a word-level transcript, because the boundary rules and two of the four scoring signals need one. Twitch VODs arrive as HLS segments rather than one file, which is slower to pull — that cost sits in the base rate now rather than in a surcharge.
Every candidate window is scored against an absolute bar
Loudness peaks, shot changes, motion, speech density, laughter and keyword matches. The score is 70 percent rank within this source and 30 percent absolute evidence, so a flat source scores like a flat source instead of being stretched to fill a nice-looking range.
The cut is resolved to sentence boundaries
The clip starts at the setup and ends after the payoff — never mid-sentence, never on a conjunction, and never before the laugh. The length you asked for is satisfied by choosing a different sentence pair, not by moving the cut to an arbitrary offset.
It is reframed, captioned and scored
9:16, 4:5, 1:1 or 16:9, with the crop following measured motion or a tracked speaker. Captions are timed word by word and kept to one clean line. Each clip ships with a virality score broken out into Hook, Flow and Ending, plus the signals behind it.
The reframe declines rather than guessing
"It follows who is talking" is two problems, and conflating them is why auto-reframe usually looks bad. Who is talking is an audio question, answered by a diarizing model. Where they are on the frame is a video question, and nothing about the audio answers it.
Momevera correlates the two: the speaker turns say when each person is talking, and the video's own per-column motion says where the picture changed while they were. Subtracting the other speakers' average column profile from this speaker's cancels everything common to both — a game feed, an animated overlay, a webcam breathing on auto-exposure — because that background is running during everyone's turns equally. What survives is the part of the frame that moves when this person talks and not when they do not.
When that contrast collapses — two people on one couch, a single-camera talking head, a shot where both speakers occupy the same columns — the pass returns nothing and the existing motion-tracked crop is kept. A held frame is a normal edit. A frame that cuts to the wrong person is a broken one. The decision is stored on the clip and shown in the editor, so a frame held because two people never separated does not look identical to a frame nobody looked at.
A real face model does run here — YuNet, over sampled frames — and its job is narrow and worth stating exactly. Diarization decides when each person speaks, the column contrast decideswhich of the people on screen that is, and the faces decide where that person actually is, to the pixel. So the crop lands on a face rather than near one.
What it is not is a mouth tracker, and it is not what picks the speaker. It is also strictly additive: when no face sits close enough to the answer, when no single face dominates, or when the render box has no detector available at all, the frame falls back to exactly the motion-tracked behaviour described above and holds. Nothing here depends on the detector being present.
What a Twitch VOD costs
- Credit unit
- 1 credit = 1 minute of source video.
- Twitch rate
- 1 credit per source minute — the same as every other source.
- Rounding
- Whole source minutes, floored, minimum 1, then multiplied and rounded up — so the quote is never below what is charged.
- Partial VOD
- Drag a processing window; you are charged for the window, not the VOD. Minimum window one minute.
- Free tier
- 90 credits a month, refills, no card, 720p exports with no watermark.
- Clip count
- Charged on source length, not on how many clips come back. A 3-hour YouTube VOD is 180 credits whether it returns 4 clips or 14.
Full plan detail is on the pricing page, including the source-length ceiling your deployment is configured with.
What comes back
- Vertical clips at 9:16, or 4:5, 1:1 and 16:9 if you want them, rendered with the source audio at 160 kbps AAC.
- Word-by-word burned-in captions, timed to the speech and kept to a single line, in a picker of caption style presets with bundled fonts — plus your own font uploads.
- A virality score with Hook, Flow and Ending broken out, and the keywords and signals that pushed each clip up the ranking.
- The clip's own opening and closing sentence, quoted, so you can see where the cut landed before you watch it.
- SRT and VTT subtitle files, and an XML export you can open in Premiere or DaVinci for final polish.
- A post kit per clip: one caption and tag list, reshaped per platform under rules written once — 150 characters for TikTok, Title Case and #Shorts for the title box, tags on their own line for Instagram, 280 including tags for X, and never #fyp on LinkedIn.
Before you paste anything
You need the right to use the video you submit. Clipping a stream or VOD you do not own may infringe someone's copyright or breach the source platform's terms; Momevera does not check this and grants you no rights in third-party content. That is spelled out in the terms, and what happens to your video is spelled out in the privacy page.
Questions, answered
Do I have to download the VOD first?
No. Paste the twitch.tv URL and Momevera fetches it. twitch.tv, www.twitch.tv and m.twitch.tv links are accepted; the host is matched exactly, so a lookalike domain is rejected rather than fetched. If you would rather upload the file yourself, that works too, at exactly the same rate — 1 credit per source minute, the same as a Twitch link and every other source.
Are Twitch links locked behind a paid plan?
No. Twitch and Kick links work on the free tier, same as YouTube, and at the same rate: 1 credit per source minute everywhere. They used to cost 1.5x, which meant a trial could not cover a single ordinary Twitch VOD -- that surcharge is gone.
Do Twitch VODs cost more credits than YouTube?
No — they cost exactly the same: 1 credit per whole source minute, with a one-credit minimum, on YouTube, Twitch, Kick, Vimeo and direct uploads alike. They are genuinely more expensive for us to fetch: a YouTube VOD is a single well-served progressive file, while a Twitch or Kick VOD is HLS — many thousands of segments pulled individually, routinely hours long, and far more likely to stall partway and need retrying. That used to be passed on as a 1.5x rate. It no longer is, because the surcharge fell entirely on the streamers this product is built for, and it was measured to be the single largest reason trial accounts could not afford their own archive.
Can I clip just the last hour of a six-hour stream?
Yes. Drag a processing timeframe over the part you care about and that is all you are charged for. The shortest window is one minute. A three-hour VOD where only the last 40 minutes matter costs 40 minutes of credits, not 180.
Does it handle a facecam and gameplay at once?
Yes — the Gameplay layout puts the facecam on top and a tracked crop of the gameplay below, and you can drag the cam box in the editor. There are also Fill (one crop that follows the action), Fit (the whole shot on a blurred bed of itself), Split (two speakers stacked) and Screen share layouts.
What resolution do the clips export at?
The free tier exports 720p, with no watermark. Creator monthly goes to 1080p, and Creator yearly and Lifetime go up to 4K. Quality is always capped by the source — a 1080p VOD cannot honestly become 4K and is never upscaled to pretend otherwise.