https://github.com/venuv/langchain_yt_tools

Langchain tools to search/extract/transcribe text transcripts of Youtube videos. Some of this has been integrated into LangChain main branch
https://github.com/venuv/langchain_yt_tools

Last synced: about 1 month ago
JSON representation

Langchain tools to search/extract/transcribe text transcripts of Youtube videos. Some of this has been integrated into LangChain main branch

• Host: GitHub
• URL: https://github.com/venuv/langchain_yt_tools
• Owner: venuv
• License: gpl-3.0
• Created: 2023-05-02T17:25:03.000Z (about 1 year ago)
• Default Branch: main
• Last Pushed: 2023-06-28T15:54:33.000Z (almost 1 year ago)
• Last Synced: 2024-05-08T00:21:30.732Z (about 2 months ago)
• Language: Python
• Homepage:
• Size: 27.3 KB
• Stars: 53
• Watchers: 2
• Forks: 4
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Lists

• awesome-langchain - langchain_yt_tools
• awesome-langchain-zh - langchain_yt_tools

README

        
# langchain_yt_tools

Custom Langchain tools to search/extract/transcribe Youtube videos

# Installation

pip install -r requirements.txt

# Code

Consists of two custom LangChain tools, _CustomYTSearchTool_ and _CustomYTTranscribeTool_

 * CustomYTSearchTool searches for the youtube videos featuring a *person name* and returns the top *num_results* URLs

 * CustomYTTranscribeTool transcribes the videos and outputs the text transcripts to transcripts.json

# Demo

running *python yt_tools.py* shows you an agent run that transcribes the top 3 video results for Laszlo Bock (Google HR celebrity). you should see the results stored in *transcripts.json* in the current working directory.

# Caveats

you will be required to authenticate with your Youtube credentials the first time around

# To Do

 * clean up the mp4 files that the Transcribe tool creates between runs

 * add threading/async capability to transcribe videos concurrently (may require not outputting transcripts to a single JSON file)