Background Music Transcription

This repository provides an automated pipeline to extract background music from YouTube videos and convert it into a textual transcription. The solution leverages widely used tools—yt-dlp for downloading and converting video audio to MP3, and Whisper for transcribing the audio content.

Prerequisites

Before using the scripts in this repository, make sure you have the following installed and set up:

• yt-dlp:
  yt-dlp GitHub Repository
  Used for downloading and converting YouTube video audio.

• FFmpeg:
  Required by yt-dlp for audio extraction:

  sudo apt update && sudo apt install ffmpeg -y

• Whisper: Whisper GitHub Repository

  pip install git+https://github.com/openai/whisper.git

Key Features

• YouTube Audio Extraction:
  Download audio directly from any YouTube URL and convert it into an MP3 file using yt-dlp.

• Automatic MP3 Detection:
  A built-in mechanism automatically searches your working directory for the downloaded MP3 file, ensuring a seamless transition to transcription.

• Audio-to-Text Transcription:
  Convert the extracted MP3 file into text using the Whisper transcription model. You can adjust the model size (e.g., small, medium, large) to balance speed and accuracy.

• Interactive Google Colab Notebook:
  For those who prefer an interactive environment, a complete Colab notebook is provided. This notebook installs the necessary dependencies and guides you through the process from audio extraction to transcription.