https://github.com/ccoreilly/datapipe

An audio ETL pipeline for generating datasets from youtube sources
https://github.com/ccoreilly/datapipe

Last synced: 12 months ago
JSON representation

An audio ETL pipeline for generating datasets from youtube sources

• Host: GitHub
• URL: https://github.com/ccoreilly/datapipe
• Owner: ccoreilly
• License: agpl-3.0
• Created: 2022-01-13T17:31:17.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2022-05-05T13:20:13.000Z (about 4 years ago)
• Last Synced: 2025-06-26T12:02:40.767Z (about 1 year ago)
• Language: Python
• Homepage:
• Size: 16.2 MB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE
  • Citation: CITATION.cff

Awesome Lists containing this project

README

          
# Audio datapipe

A data processing pipeline that (currently) extracts audio clips from youtube videos and generates two transcription candidates with a Vosk (Kaldi) and a Wav2Vec2 model. The goal of the software is to ease the generation of datasets for ASR by automatically extracting and processing large audio sources.

This software has been written for internal use and made public as it might be useful for someone. However, no support and no warranty will be provided.

Copyright © 2021 Ciaran O'Reilly

Licensed under the GNU AFFERO GENERAL PUBLIC LICENSE (AGPLv3)