video-add-chapters_skill

This skill transcribes videos, detects chapters, and generates YouTube-ready markdown with optional highlight videos to streamline documentation.

Python

24

GitHub Stars

9

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill jykim/claude-obsidian-skills --skill video-add-chapters

clean_transcript.py8.3 KB
export_highlight_script.py4.6 KB
generate_docs.py10.4 KB
generate_highlights.py7.5 KB
parse_highlight_annotations.py7.1 KB
requirements.txt14 B
SKILL.md9.6 KB
suggest_chapters.py8.6 KB
transcribe_video.py7.7 KB

Overview

This skill adds chapters to videos by transcribing audio, detecting topic boundaries, and producing structured markdown documents with YouTube chapter markers. It uses Whisper for transcription, analyzes the transcript to suggest chapter boundaries, and generates per-chapter markdown plus a merged document and chapter index. Optionally, it can produce highlight videos from selected segments.

How this skill works

The pipeline extracts audio and transcribes in 15-minute chunks via the Whisper API, preserving timestamps and handling offsets. It analyzes the transcript for transition signals to suggest chapter boundaries, then generates individual chapter files, a merged document, and a YouTube chapter list. For highlights, it parses annotated transcript segments or an editable script to create FFmpeg commands and assemble a highlights video with optional titles and padding.

When to use it

Transcribing long videos (20+ minutes) and splitting into logical chapters
Converting raw transcripts into clean, readable documentation
Generating YouTube chapter markers for video descriptions
Creating highlight videos from selected transcript segments
Cleaning filler words and improving transcript cohesion

Best practices

Transcribe once and reuse transcripts with --skip-if-exists to save API cost
Review suggested chapter.json and adjust timestamps before final document generation if transitions are ambiguous
Use <u> or == annotations in the transcript to mark highlight segments for precise highlight generation
Run clean_transcript.py after chapter detection to preserve timestamps while improving readability
Verify chapter start times against the source video after automated detection

Example use cases

Produce a chaptered markdown manual for long tutorial recordings and publish YouTube chapter markers
Create a highlights reel by annotating key transcript lines and exporting a merged highlight video
Convert meeting recordings into searchable chaptered notes for team knowledge bases
Generate an index and per-chapter markdown files for documentation or course materials
Clean and republish transcripts with timestamps preserved for accessibility

FAQ

The pipeline is optimized for Korean currently; multi-language support is planned but not yet implemented.

How are highlight segments defined?

Highlights are defined via in-transcript annotations (<u> or ==) or by exporting/editing a highlight script with [START-END] timestamps and optional {titles}.

What tools are required?

Python 3.7+, FFmpeg for audio/video processing, and an OPENAI_API_KEY for Whisper transcription.