Showcase: mr-transcript – a convenient Python wrapper for easier YouTube transcript retrieval

[BogdanovychA]

Showcase: mr-transcript – a convenient Python wrapper for easier YouTube transcript retrieval

Hi everyone!

I'd like to share a small open-source tool I've built on top of the excellent youtube-transcript-api library.

It's called Mr. Transcript.

Why I built it

While youtube-transcript-api is a powerful tool, many projects (especially when working with AI agents or quick scripts) often require the same boilerplate code for:

1. URL Parsing: Extracting the video_id from various formats (shorts, mobile, embed, youtu.be, etc.).
2. Priority Logic: Attempting to fetch manually created transcripts first, and falling back to auto-generated ones only if necessary.
3. Formatting: Converting the list of dictionaries into a clean string (with or without timestamps).

Key Features:

• Universal Link Support: Supports almost all YouTube link formats out of the box.
• Smart Source Selection: Automatically prioritizes manual transcripts over generated ones.
• Easy Timestamp Management: The timecodes=True option instantly adds time markers to every line.
• Type Safety: Full support for type annotations for a better developer experience.
• AI-Ready: Includes specialized "skills" for AI agents, making it easier to integrate the library into LLM pipelines.

Quick Example:

from mr_transcript import get_transcript

# Works with both IDs and full links (including Shorts)
url = "https://www.youtube.com/watch?v=..."

# Fetch the ready-to-use text in one call
text = get_transcript(url, language="en", timecodes=True)
print(text[:300])

I'd love to hear your thoughts or feedback! I hope this wrapper proves useful for anyone looking for the simplest way to integrate transcripts into their Python projects.

Repository: https://github.com/BogdanovychA/mr-transcript

[jdepoix]

Hi!

not to take anything away from what you've built, but I just wanted to point out that both YouTubeTranscriptApi.fetch and TranscriptList.find_transcript already prioritize manually generated transcripts over generated ones for the requested language!