Production-ready animations, transitions, backgrounds, and scenes for Remotion
TypeScript ai-agentsclaude-codedemo-videomcpmcp-servermodel-context-protocolmotion-designproduct-demoreactremotionshadcnvideo
{
"$schema": "https://ui.shadcn.com/schema/registry-item.json",
"name": "caption-core",
"title": "Caption Core",
"description": "The timing engine every caption style shares: pages a timed transcript, finds the word being spoken and how far into it the voice is, and decides when a page leaves the screen.",
"dependencies": ["remotion", "@remotion/captions"],
"files": [
{
"path": "registry/remocn/caption-core/index.ts",
"content": "import {\n type Caption,\n createTikTokStyleCaptions,\n type TikTokPage,\n} from \"@remotion/captions\";\nimport { useLayoutEffect, useMemo, useRef, useState } from \"react\";\nimport { useCurrentFrame, useVideoConfig } from \"remotion\";\n\nexport type SpeakerCaption = Caption & { speaker?: string };\n\nexport interface CaptionBaseProps {\n captions: Caption[];\n combineTokensWithinMilliseconds?: number;\n holdMs?: number;\n fontSize?: number;\n fontWeight?: number;\n color?: string;\n className?: string;\n}\n\nexport const DEFAULT_HOLD_MS = 600;\nexport const MIN_SILENCE_BREAK_MS = 300;\n\nexport interface CaptionPageWindow {\n page: TikTokPage;\n index: number;\n speaker: string | undefined;\n startMs: number;\n speechEndMs: number;\n endMs: number;\n endsInSilence: boolean;\n}\n\nexport type CaptionTokenStatus = \"upcoming\" | \"active\" | \"spoken\";\n\nexport interface CaptionTokenState {\n text: string;\n spaceBefore: boolean;\n fromMs: number;\n toMs: number;\n progress: number;\n status: CaptionTokenStatus;\n}\n\nexport interface CaptionPreviousPage {\n page: TikTokPage;\n pageIndex: number;\n tokens: CaptionTokenState[];\n}\n\nexport interface CaptionsState {\n page: TikTokPage;\n pageIndex: number;\n speaker: string | undefined;\n timeMs: number;\n activeIndex: number;\n tokens: CaptionTokenState[];\n sinceStartMs: number;\n untilEndMs: number;\n holdRoomMs: number;\n endsInSilence: boolean;\n previous: CaptionPreviousPage | null;\n}\n\nexport interface CaptionPagingOptions {\n combineTokensWithinMilliseconds: number;\n holdMs?: number;\n}\n\nconst SENTENCE_END = /[.!?…][\"')\\]]*$/;\n\nfunction markPageBreaks(captions: SpeakerCaption[]): Caption[] {\n return captions.map((caption, i) => {\n if (caption.pageBreakAfter !== undefined) return caption;\n const next = captions[i + 1];\n const speakerChanges =\n next !== undefined && next.speaker !== caption.speaker;\n return speakerChanges || SENTENCE_END.test(caption.text.trim())\n ? { ...caption, pageBreakAfter: true }\n : caption;\n });\n}\n\nexport function buildCaptionPages(\n captions: SpeakerCaption[],\n {\n combineTokensWithinMilliseconds,\n holdMs = DEFAULT_HOLD_MS,\n }: CaptionPagingOptions,\n): CaptionPageWindow[] {\n if (captions.length === 0) return [];\n const { pages } = createTikTokStyleCaptions({\n captions: markPageBreaks(captions),\n combineTokensWithinMilliseconds,\n breakOnSilenceAfterMilliseconds: Math.max(holdMs, MIN_SILENCE_BREAK_MS),\n });\n const speakers = new Map<number, string | undefined>();\n for (const caption of captions) {\n if (!speakers.has(caption.startMs)) {\n speakers.set(caption.startMs, caption.speaker);\n }\n }\n const spoken = pages.filter((page) => page.tokens.length > 0);\n return spoken.map((page, index) => {\n const last = page.tokens[page.tokens.length - 1];\n const speechEndMs = last ? last.toMs : page.startMs;\n const heldUntil = speechEndMs + Math.max(0, holdMs);\n const next = spoken[index + 1];\n const endsInSilence = !next || next.startMs > heldUntil;\n const endMs = endsInSilence\n ? heldUntil\n : Math.max(page.startMs, next.startMs);\n return {\n page,\n index,\n speaker: speakers.get(page.tokens[0].fromMs),\n startMs: page.startMs,\n speechEndMs,\n endMs,\n endsInSilence,\n };\n });\n}\n\nfunction findWindowIndex(windows: CaptionPageWindow[], timeMs: number): number {\n let lo = 0;\n let hi = windows.length - 1;\n let found = -1;\n while (lo <= hi) {\n const mid = (lo + hi) >> 1;\n if (windows[mid].startMs <= timeMs) {\n found = mid;\n lo = mid + 1;\n } else {\n hi = mid - 1;\n }\n }\n if (found === -1) return -1;\n return timeMs < windows[found].endMs ? found : -1;\n}\n\nfunction tokenProgress(fromMs: number, toMs: number, timeMs: number): number {\n if (timeMs < fromMs) return 0;\n if (toMs <= fromMs) return 1;\n return Math.min(1, (timeMs - fromMs) / (toMs - fromMs));\n}\n\nfunction getTokenStates(\n page: TikTokPage,\n timeMs: number,\n): { activeIndex: number; tokens: CaptionTokenState[] } {\n let activeIndex = -1;\n page.tokens.forEach((token, i) => {\n if (token.fromMs <= timeMs) activeIndex = i;\n });\n const tokens = page.tokens.map(\n (token, i): CaptionTokenState => ({\n text: token.text.trim(),\n spaceBefore: i > 0 && /^\\s/.test(token.text),\n fromMs: token.fromMs,\n toMs: token.toMs,\n progress: tokenProgress(token.fromMs, token.toMs, timeMs),\n status:\n i < activeIndex ? \"spoken\" : i === activeIndex ? \"active\" : \"upcoming\",\n }),\n );\n return { activeIndex, tokens };\n}\n\nexport function getCaptionsState(\n windows: CaptionPageWindow[],\n timeMs: number,\n): CaptionsState | null {\n const found = findWindowIndex(windows, timeMs);\n if (found === -1) return null;\n const window = windows[found];\n const { activeIndex, tokens } = getTokenStates(window.page, timeMs);\n const before = found > 0 ? windows[found - 1] : null;\n const previous =\n before && !before.endsInSilence\n ? {\n page: before.page,\n pageIndex: before.index,\n tokens: getTokenStates(before.page, before.endMs).tokens,\n }\n : null;\n return {\n page: window.page,\n pageIndex: window.index,\n speaker: window.speaker,\n timeMs,\n activeIndex,\n tokens,\n sinceStartMs: timeMs - window.startMs,\n untilEndMs: window.endMs - timeMs,\n holdRoomMs: window.endMs - window.speechEndMs,\n endsInSilence: window.endsInSilence,\n previous,\n };\n}\n\nexport function getExitWindowMs(state: CaptionsState, exitMs: number): number {\n return Math.max(1, Math.min(exitMs, state.holdRoomMs));\n}\n\nexport function groupCaptionWords(\n tokens: CaptionTokenState[],\n): { index: number; token: CaptionTokenState }[][] {\n const words: { index: number; token: CaptionTokenState }[][] = [];\n tokens.forEach((token, index) => {\n const current = words[words.length - 1];\n if (!current || token.spaceBefore) {\n words.push([{ index, token }]);\n } else {\n current.push({ index, token });\n }\n });\n return words;\n}\n\nexport function normalizeCaptionWord(text: string): string {\n return text\n .toLowerCase()\n .replace(/[\\u2018\\u2019]/g, \"'\")\n .replace(/[^\\p{L}\\p{N}']/gu, \"\");\n}\n\nexport function createKeywordMatcher(\n keywords: readonly string[] = [],\n): (text: string) => boolean {\n const set = new Set(keywords.map(normalizeCaptionWord).filter(Boolean));\n return (text) => set.has(normalizeCaptionWord(text));\n}\n\nexport function captionWordText(word: { token: CaptionTokenState }[]): string {\n return word.map(({ token }) => token.text).join(\"\");\n}\n\nexport interface CaptionWordRect {\n left: number;\n top: number;\n width: number;\n height: number;\n}\n\nfunction sameRects(a: CaptionWordRect[], b: CaptionWordRect[]): boolean {\n return (\n a.length === b.length &&\n a.every(\n (r, i) =>\n r.left === b[i].left &&\n r.top === b[i].top &&\n r.width === b[i].width &&\n r.height === b[i].height,\n )\n );\n}\n\nexport function useCaptionWordRects(\n pageIndex: number,\n count: number,\n): {\n wordRef: (index: number) => (el: HTMLElement | null) => void;\n rects: CaptionWordRect[] | null;\n} {\n const elements = useRef<(HTMLElement | null)[]>([]);\n const [measured, setMeasured] = useState<{\n pageIndex: number;\n rects: CaptionWordRect[];\n } | null>(null);\n\n useLayoutEffect(() => {\n if (pageIndex === -1) return;\n const rects = elements.current.slice(0, count).map((el) =>\n el\n ? {\n left: el.offsetLeft,\n top: el.offsetTop,\n width: el.offsetWidth,\n height: el.offsetHeight,\n }\n : { left: 0, top: 0, width: 0, height: 0 },\n );\n setMeasured((prev) =>\n prev && prev.pageIndex === pageIndex && sameRects(prev.rects, rects)\n ? prev\n : { pageIndex, rects },\n );\n });\n\n return {\n wordRef: (index) => (el) => {\n elements.current[index] = el;\n },\n rects: measured && measured.pageIndex === pageIndex ? measured.rects : null,\n };\n}\n\nexport function useCaptions({\n captions,\n combineTokensWithinMilliseconds,\n holdMs = DEFAULT_HOLD_MS,\n}: {\n captions: SpeakerCaption[];\n combineTokensWithinMilliseconds: number;\n holdMs?: number;\n}): CaptionsState | null {\n const frame = useCurrentFrame();\n const { fps } = useVideoConfig();\n const windows = useMemo(\n () =>\n buildCaptionPages(captions, { combineTokensWithinMilliseconds, holdMs }),\n [captions, combineTokensWithinMilliseconds, holdMs],\n );\n return getCaptionsState(windows, (frame / fps) * 1000);\n}\n",
"type": "registry:lib",
"target": "lib/remocn/caption-core.ts"
}
],
"type": "registry:lib"
}