Can I choose the transcription quality?

Yes, when using the Whisper fallback, the skill defaults to the 'base' model for a balance of speed and accuracy, but it can be configured to use other model sizes depending on your hardware.

How does it handle the messy formatting of auto-generated subtitles?

It includes a Python post-processing script that automatically removes VTT metadata and deduplicates lines that often repeat in YouTube's rolling caption format.

Does this skill work for videos without any captions?

Yes. If no subtitles are available on YouTube, the skill can download the video's audio and use the Whisper AI model to generate a new transcription locally.

What tools need to be installed for this to work?

The skill primarily uses yt-dlp for downloads and Whisper for transcription. It includes logic to check for these tools and can assist with their installation via Homebrew, apt, or pip.

YouTube Transcript Downloader

Name: YouTube Transcript Downloader
Author: lifangda

bylifangda

•

Learning & Documentation

Automates the extraction, transcription, and cleaning of YouTube video subtitles and captions into readable text.

About

This skill empowers Claude to retrieve and process text content from YouTube videos using a multi-layered approach. It first attempts to download high-quality manual or auto-generated subtitles using yt-dlp. If no captions are available, it provides a sophisticated fallback that downloads the audio and utilizes OpenAI Whisper for local AI-powered transcription. The skill includes built-in post-processing to remove VTT formatting and deduplicate overlapping lines, resulting in clean, production-ready text files optimized for reading or analysis.

Key Features

Automated download of manual and auto-generated YouTube subtitles
Automatic dependency management for yt-dlp and Whisper installations
Smart filename sanitization based on video titles
16 GitHub stars
Fallback AI transcription using Whisper for videos without native captions
Intelligent text deduplication to clean up overlapping VTT lines

Use Cases

Converting video tutorials into written documentation or blog posts
Summarizing educational videos, lectures, or conference talks
Creating searchable text databases from video archives

About

Key Features

Automated download of manual and auto-generated YouTube subtitles
Automatic dependency management for yt-dlp and Whisper installations
Smart filename sanitization based on video titles
16 GitHub stars
Fallback AI transcription using Whisper for videos without native captions
Intelligent text deduplication to clean up overlapping VTT lines

Use Cases

Converting video tutorials into written documentation or blog posts
Summarizing educational videos, lectures, or conference talks
Creating searchable text databases from video archives