Audio Analysis Setup Guide

About

This is a project to test google cloud deployment of an audio-analysis function. It can also be run locally.

Prerequisites

1. Python 3.8+ installed on your system
2. Audio files (MP3, WAV, FLAC, M4A, or AAC format)

Installation Steps

1. Create a Virtual Environment (Recommended)

python -m venv audio_analysis_env

# On Windows:
audio_analysis_env\Scripts\activate

# On macOS/Linux:
source audio_analysis_env/bin/activate

2. Install Required Packages

pip install librosa numpy pathlib

Note: librosa might take a few minutes to install as it has several dependencies including scipy, scikit-learn, and audio processing libraries.

3. Test Your Installation

Create a small test script to verify everything works:

import librosa
import numpy as np
print("All packages installed successfully!")

Running the Analysis

1. Prepare your audio files: Put some music files in a folder
2. Run the script: Execute the Python script and enter your music folder path
3. Check results: Look for the generated audio_analysis_results.json file

What the Analysis Provides

Basic Information

• Duration: Length of the track
• Tempo: Beats per minute (BPM)
• Estimated Key: Musical key (C, D, E, etc.)

Audio Features

• Spectral Centroid: Brightness indicator (higher = brighter sound)
• RMS Energy: Overall loudness/energy level
• Zero Crossing Rate: Measure of how "noisy" vs "tonal" the audio is
• MFCCs: Audio fingerprint (useful for similarity matching later)

Mood Indicators

• Energy Level: High/Medium/Low based on tempo and RMS
• Brightness: Bright/Balanced/Dark based on spectral content
• Rhythmic Stability: How consistent the rhythm is

Example Output

{
  "file_name": "example_song.mp3",
  "duration_seconds": 240.5,
  "tempo_bpm": 128.4,
  "estimated_key": "C",
  "mood_indicators": {
    "energy_level": "High Energy",
    "brightness": "Bright",
    "rhythmic_stability": "Stable"
  }
}

Start with 3-5 songs from different genres to see how the analysis varies!