Captions that stay word-timed, from the same transcript you already paid for.
5 presets: Impact Anton, Clean Inter, Block Archivo, Story Poppins, Space Grotesk — plus accent, scale, baseline, punch words, 9:16/1:1/16:9. Full-stream SRT/VTT download is free and reuses the same word-timed chunks that power 20/h suggestions. Change preset without re-transcribing.
Most caption tools re-transcribe for every export. Weaver transcribes once, then renders any preset from the same words. A 45s clip with 'pixel' punched is 3 cards: leadIn 100ms, hold 150ms, min 400ms, max 16 chars for 9:16, 34 for 16:9, baseline 0.74. Change from Impact to Space and the words don't move — only the font, stroke, tracking, and accent do. That's why a transcript hour is the scarcest resource and captions are free to iterate.
5 presets: Impact (Anton, yellow), Clean (Inter, scale), Blo
Word-timed: every word has startMs/endMs, punch words, accen
Full-stream SRT/VTT stitched from same chunks — download fro
Safe-Zone Previewer: see webcam/chat/caption zones before yo
SRT ⇄ VTT ⇄ TXT, overlap fixer, 2-line linter
All in browser — your file never leaves the page.
See the 3 moments we'd pull before you sign up
MCP weaverclip_search_transcript — free taste.
How much silence are you transcribing?
Is your old VOD already gone?
Twitch deletes fast. WeaverClip keeps it.
Does your lane keep up live?
Record bitrate vs upstream — will you finish uploading after you stop?
Two numbers match or you don't touch the file
MCP weaverclip_get_usage — verify before delete.
One canvas or two?
1.4 TB freed — zero files lost
Will your webcam get cropped?
Do people actually read this?
*Hours are at 6 Mbps (1080p30 webinar, slides, sermons) ≈2.7 GB/hour. Actual hours vary with bitrate — 5 Mbps ~2.25 GB/h, 12 Mbps ~5.4 GB/h, 25 Mbps ~11.25 GB/h. Vault is always bytes; hour label is a guide. Hard limit = 2× vault (e.g., Creator 75→150 GB), overage $0.04/GB-month. Transcript never bills overage.
Sharp nuances you didn't know you needed
How word-timed captions work — cards, not blobs
CaptionWords: text, startMs, endMs, punch. Cards: 2–3 words, max 16 chars for 9:16, 34 for 16:9, leadIn 100ms, hold 150ms, min 400ms, baseline 0.74, actionRail 0.16. A 10-word sentence becomes 4 cards, each ~1.2s. The renderer builds CaptionCards from words, not sentences, so 'burnout was not the workload' becomes 3 cards with punch on 'burnout'. Change preset and the cards rebuild from same words — no re-transcribe, no extra the transcript allowance.
The 5 presets — why they exist and when to use which
Impact (Anton, upper, heavy stroke #FFD400) for gaming hype — 1.4× scale on punch. Clean (Inter Black, sentence, scale) for tech commentary — restrained shadow. Block (Archivo Black, upper, block #FF5A36) for reactions — active block behind word. Story (Poppins SemiBold, sentence, accent #F0BFAF) for podcasts — soft. Space (Space Grotesk, sentence, accent #7CC6F5, wide tracking 0.015) for interviews/essays — airy, modern, sky. BestFor is not marketing — it's the stroke width and tracking that makes Space readable at 9:16 with 2 lines.
Safe-Zone Previewer — the 9:16 mistake you make once
Upload a 16:9 frame, see webcam 40px outside the vertical crop, move it before you record 6h. The previewer counts against your 2 free layouts (Free) or 30/60 (paid), first regen free. It renders a matched 16:9 + 9:16 plate pair so you see both canvases at once. Biggest streamers care because their webcam is 6 hours of brand — one Safe-Zone check saves 6 hours of 'why is my face cut off on TikTok?'.
Full-stream SRT/VTT — the upload-once, clip-later unlock
From any session with transcript, FullStreamCaptions stitches transcript_chunks into SRT: 1\n00:00:01,200 --> 00:00:03,400\nText and VTT: WEBVTT\n00:00:01.200 -->. Same words that power 20/h suggestions. Download tonight, import to Descript/Premiere, come back next week to bulk-create 18 edits via weaverclip_bulk_create_edits without re-transcribing. That's the 'plus the clips later' you asked for — upload 24h once, get .srt now, clips later.
Caption readability — WPM, chars/line, punch discipline
Caption Readability tool on /features/captions: paste transcript + duration → WPM, chars/line, 'needs 2 lines' flags. You talk at 187 WPM, your captions need two lines or nobody reads them. Punch words 1–2 per clip, more than that and none land. The tool uses the same maxCharactersPerCard (16 for 9:16) the renderer uses, so the preview is honest, not decoration.
Questions that decide the purchase — answered before you ask
Do I need to transcribe again to change preset?
No. Preset is rendering, not transcription. Change from Clean to Space and the helper rebuilds cards from same words — same startMs/endMs, new font, same transcript hour. That's why transcript counts once and captions are free to iterate. Only new source minutes that never had transcript cost.
Which preset for which streamer?
Gaming/Valorant → Impact (yellow, heavy stroke, 1.4× punch on 'pixel'). Tech commentary → Clean (Inter, scale). IRL reactions → Block (coral block). Podcasts/teaching → Story (blush) or Space (sky, wide tracking for interviews). Biggest streamers pick Space for 3-hour podcasts because wide tracking at 0.015 is readable on phone without eye strain.
Can I download the full transcript as SRT without making a clip?
Yes. Every session with transcript shows FullStreamCaptions on /sessions/[id] — SRT and VTT buttons above the transcript. It's stitched from transcript_chunks, not from clip captions. One download, all cues, word-timed, ready for YouTube chapters or Descript.
What about 9:16 safe zones for dual canvas?
The previewer renders both 16:9 and 9:16 from one plate pair so you see the crop before you record. Webcam, chat, captions each have zones; the preview marks them. Counts against 2 free layouts, first regen free. Most 6-hour 'why is my face cut off' mistakes are fixed in 30 seconds here.
Try it with the video you already have.
Free holds a full 2–3h webinar at 6 Mbps for 30 days. Upload the whole thing, download the SRT, make clips later — no re-transcribe.