youtube-transcript

name

youtube-transcript

description

Extract transcripts from YouTube videos, local audio/video files, and summarize transcript files. Use when the user asks to download, fetch, extract, transcribe, caption, or summarize a YouTube video, a YouTube URL, a video ID, a local media file such as .mp4/.mov/.mp3/.m4a/.wav, or an existing transcript/text file. Supports captions-only, chunked local media transcription through OpenAI, YouTube audio fallback through OpenAI, and summaries through OpenAI or opt-in Ollama.

YouTube Transcript

Use the bundled script for deterministic transcript work instead of recreating the workflow.

Quick Start

Run from the skill directory or call the script by absolute path:

python scripts\yt_transcribe.py "https://www.youtube.com/watch?v=VIDEO_ID"

Default behavior for YouTube input:

• Try YouTube captions first with youtube-transcript-api.
• If captions fail, download audio with yt-dlp and transcribe with OpenAI.
• Write .transcript.md and .metadata.json into ~/Documents/YouTube Transcripts.
• Do not summarize unless --summary is supplied.
• Use --output-dir "path\to\folder" when the user asks for a specific output folder. Prefer a user data folder, not a git repo checkout.

For local audio/video files, pass the file path directly. The script extracts audio with ffmpeg, using bundled imageio-ffmpeg if no system ffmpeg is installed, chunks it, and transcribes each chunk with OpenAI.

Common Tasks

Captions only, no audio fallback:

python scripts\yt_transcribe.py "https://www.youtube.com/watch?v=VIDEO_ID" --captions-only

Custom output folder:

python scripts\yt_transcribe.py "https://www.youtube.com/watch?v=VIDEO_ID" --output-dir "C:\Users\me\Documents\Transcripts"

Transcript plus OpenAI summary:

python scripts\yt_transcribe.py "https://www.youtube.com/watch?v=VIDEO_ID" --summary

Summarize an existing transcript/text file:

python scripts\yt_transcribe.py --file "path\to\transcript.md" --summary

Force audio transcription:

python scripts\yt_transcribe.py "https://www.youtube.com/watch?v=VIDEO_ID" --audio-only

Transcribe a local video or audio file:

python scripts\yt_transcribe.py "C:\Users\me\Downloads\talk.mp4"

API Keys

Prefer a .env file in the current working directory:

OPENAI_API_KEY=sk-...

The script also accepts OPENAI_API_KEY, OPENAI_KEY, OPENAI_API_TOKEN, or --openai-api-key. Avoid putting keys in shell history unless the user explicitly chooses that.

Notes

• Use --captions-only for long videos when the user wants the fast/simple path.
• OpenAI is the default summarizer. Use --summarizer ollama only when the user explicitly wants local Ollama.
• The script self-installs missing Python packages with pip install --upgrade: youtube-transcript-api, yt-dlp, openai, requests, and imageio-ffmpeg.
• For large transcripts, OpenAI summaries are chunked automatically.