Write a video as Markdown or JSON. makevideo voices it with a local neural voice, times every scene to the real audio, adds captions and transitions, and renders an MP4 — on a plain CPU, faster than it plays.
# video.md
# Weekly update
What shipped this week
> Here is what shipped this week.
## Highlights
- Faster builds
- New onboarding
> Two highlights. Builds are faster. And onboarding is new.npx -y github:random00000000/makevideo video.md # → video.mp4 + video.srt + poster + report
npx -y github:random00000000/makevideo video.md --fast # ~10x real-time voice for drafts
npx -y github:random00000000/makevideo preview video.md # contact sheet of every scene in ~1 sClaude Desktop, Cursor, VS Code, Codex: add npx -y github:random00000000/makevideo mcp. Tools: make_video, preview_video (returns an image), check_video.
/plugin marketplace add random00000000/makevideo — then just say “make a video about…”.
npx -y github:random00000000/makevideo agents --all writes a “use makevideo for videos” block into AGENTS.md, CLAUDE.md, Cursor rules and Copilot instructions.
Every scene is spoken. Same voice on every video. Symbols like 42%, v1.2 and $3M are read correctly.
Transitions blended from real frames, kinetic statements, animated charts, split media layouts, bundled fonts.
Lint flags walls of text and unsynced lists before rendering. Errors name the scene and show a working example.
Frame holding, render processes and cached speech: a 74 s video re-renders in ~28 s on a laptop.
Record gameplay from a window, then turn clips into a narrated trailer or devlog with titles, captions and transitions — all from the command line.
A repeatable way for coding agents to produce narrated, captioned MP4 videos — explainers, recaps, changelogs, demos — without building a new pipeline every time.
An honest comparison of ways to generate video from code, and when an opinionated, narration-first tool like makevideo is the right choice.
Write a video as Markdown — headings are scenes, quote lines are narration — and render it to a narrated, captioned MP4 with one command.
Run makevideo as a Model Context Protocol server so any MCP-capable assistant can preview and render narrated MP4 videos.
Generate narrated MP4 videos from text on a plain CPU in Node.js, with a local neural TTS voice, captions and transitions. Measured timings included.
Generate vertical 1080x1920 short-form videos with big word-by-word captions, a local voice-over and GIF export — from Markdown or JSON, on a CPU.