← ClaudeAtlas

youtube-english-subtitle-extractorlisted

Find a YouTube video and extract its English transcript or burned-in captions; when neither is available, locally transcribe Urdu audio with the specified Whisper model.
saad-git-007/claude-skills · ★ 1 · Data & Documents · score 74
Install: claude install-skill saad-git-007/claude-skills
# YouTube English Subtitle Extractor Use this skill when the user gives a YouTube video title or link and wants its English subtitles or transcript extracted. Search for the video, confirm the intended match from title/channel/duration, and produce a local, timestamped transcript artifact. ## Workflow 1. **Find the right video.** Search YouTube using the supplied title (or resolve the supplied link). If multiple plausible matches remain, show the candidates and ask which one before downloading. Use the user's stated channel, date, or other identifying details when available. 2. **Check for an English transcript before OCR.** First inspect YouTube subtitle tracks with `yt-dlp` (`--list-subs`), including manually provided and auto-generated tracks. If needed, use the available browser-use/browser tools to open the video, expand its description if necessary, select **Show transcript**, and inspect the transcript language. A Bangla, Urdu, or other non-English transcript does not substitute for English captions burned into the image. 3. **Download the video.** By default, save a local copy at a resolution sufficient to read the captions (720p is usually enough). If the user explicitly asks only for a transcript or captions and not a video copy, avoid downloading the video unless needed to access the transcript. If the video is unavailable or access-restricted, report that instead of trying to bypass the restriction. 4. **If English captions/transcript are available, download th