YouTube Audio Analyzer

Purpose

This project was written to test audio analysis methods in Python. To create data to work with, it starts by scraping the audio from YouTube videos. This is a work in progress. At this point, we can scrape audio from the most popular videos in different YouTube video categories. Audio mining yet to be completed.

First time setup

1. The Python package used to scrape audio from YouTube videos (youtube-dl) has software dependencies. We used these steps to get set up:

   • Download youtube-dl from: https://yt-dl.org/
   • Define youtube-dl.exe's location in system PATH environment variable.
   • Download LIBAV from this link and copy the .exe and all DLL files to the location of the youtube-dl.exe: http://builds.libav.org/windows/release-gpl/

2. Set up a YouTube API key at: https://console.cloud.google.com/

3. After getting software dependencies and an API key taken care of, we stored our API key in a .env file. Replace the file path in the load_dotenv() command to your own .env location in the download_audio_from_youtube.py file to run it.

4. Install Python dependencies with pip install -r requirements.txt

To do

• Use PyAudioAnalysis to extract features from audio files
• Use features for classification