Document To Narration

This converts written documents into TTS narration with word-level timing for video production. It splits your markdown into natural scene breaks based on argument flow rather than just heading structure, generates audio using a local Qwen3-TTS model, then transcribes with Whisper to get precise word timestamps for captions. The full narration pipeline is smarter than the legacy per-scene approach because it avoids volume jumps between concatenated clips. You get scene text files, WAV audio, VTT captions, and timing boundaries ready for video composition. Requires a chunky local setup with Deno, Python 3.12, ffmpeg, and a 7.8GB TTS model, so this is for people already committed to local TTS workflows who need that granular timing data.

skills.sh installs: 994
GitHub stars: 182
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add jwynia/agent-skills --skill document-to-narration --agent claude-code

Installs into .claude/skills of the current project.

Files

Document to Narration

Convert written documents into narrated video scripts with precise word-level timing.

Core Principle

The agent interprets; the document guides. Rather than rigid template-based splits, this skill uses agent judgment to find where the content naturally breathes, argues, and transitions. The document's argument flow determines scene breaks, not a predetermined structure.