yttran

yttran (“YouTube Transcript”) is a lightweight CLI tool that allows you to quickly fetch the plain transcript text from YouTube videos without downloading the video itself. It leverages the official youtube-transcript-api and can be run via uvx for easy, one-step usage.

Problem

YouTube videos often have captions or auto-generated transcripts.

There are websites like https://transcribefromyoutube.com/ that one can use to get a transcript but they require you to leave the terminal and copy paste the text for further processing.

A CLI lternative is to use the cli that comes with youtube-transcript-api but I can never remember all the switches nor do I want to type all of those long commands.

uvx run youtube_transcript_api <first_video_id> <second_video_id> ... --languages en --format text > transcripts.txt

I need to be able to run a short command and specify only the youtube video so that I can pipe transcript into AI analysis and summarization tools. I also what to save the transacript into a local cache to I don't need to fetch it again.

Further processing to extract summaries and wisdom can be done with tools like Claude AI, ChatGPT or Fabric.

Insummary, in order to maximize learning from Youtube videos...

• Downloading full videos just to get transcription text is cumbersome.
• Subtitles come in .srt or .vtt formats, which include timestamps, numbering, and formatting tags.
• Cleaning and deduplicating these files is tedious, especially when automating workflows.

yttran solves this by providing a single command to get clean, readable transcript text, ready for passing on to AI pipelines, research, or note-taking.

Dependency

uv manages project dependencies and environments for python packages.

See https://docs.astral.sh/uv/#installation

Code repository

https://github.com/rnwolf/yttran

Install package during development for testing

uv pip install .

Now you can run the cli tool binary.

Example run on MS-Windows:

.\.venv\Scripts\yttran.exe https://www.youtube.com/watch?v=4b0iet22VIk > test-tran.txt

Publish Packge to PyPi for world to use

Make sure verions number of package is incremented as required in pyproject.toml

uv build
uv publish

You’ll need your PyPI credentials (username and password).

After publishing, verify your package is live on PyPI by visiting:

https://test.pypi.org/project/yttran/

Installation Published Public Package for use on local PC

If you have uvx installed, you can run yttan directly from PyPI without manual dependency management:

uvx yttran <youtube-video-url-or-id> > transcript.txt