Learn to preprocess an audio dataset for music genre classification in this 38-minute tutorial. Implement code to batch process the Marsyas music dataset, extracting MFCCs and genre labels. Save the data in a JSON file format optimized for classifier training. Access the provided GitHub repository for the complete code and find the Marsyas genre dataset on Kaggle. Explore topics including dataset introduction, preprocessor setup, dictionary creation, file path handling, semantic label management, audio file loading, sample segmentation, and MFCC vector calculation. Conclude with a demonstration of the preprocessing results and gain practical insights into preparing audio data for machine learning applications.
Overview
Syllabus
Introduction
Music dataset
Preprocessor
Dictionary
Walk
Count
Path
Save semantic label
Append semantic label
Load audio file
Sample per segment
Samples per track
Expected number of M FCC vectors
Seal function
Print data
Run function
Results
Taught by
Valerio Velardo - The Sound of AI
