🎶 Welcome to the EDM genre classification dataset! 🎶

Take a trip through the colorful world of music with our carefully put together dataset! 🎶. This dataset got 16 different music styles for you to dive into. Whether you're into the chill vibes of Ambient 🌌 or the energetic beats of Big Room House 🏠, there's something for everyone here. Each genre brings its own special sound, just waiting for you to discover. 🎵

List of genres:

• Ambient 🌌
• Big Room House 🏠
• Drum and Bass 🥁
• Dubstep 🎵
• Future Garage/Wave Trap 🌊
• Hardcore 🔊
• Hardstyle 💥
• House 🏡
• Lo-fi 🎶
• Moombahton/Reggaeton 🎵🌴
• Phonk 🔥
• Psytrance 🌀
• Synthwave 🎹
• Techno 🎛️
• Trance 🚀
• Trap ⛓️

Each one of these genres and represented equally across both the training and testing dataset.

How this dataset was curated?

All audio clips used for feature extraction are sourced from YouTube music mixes, ensuring a diverse and extensive collection of musical content across different genres. Then the music files are loaded up into "Ableton" and sliced into equal length 3 second audio clips. These audio clips are used to extract the features.

Inorder the keep the dataset consistent. I equalized the number of entries per genre 2500 and it is split into 2000 training features per genre and 500 testing features per genre. This data-splitting is done before feature extraction so I can guarantee that there is no data leakage, however since the music are taken from splitting larger excerpts, some rows of data might be similar/same and not all 3 second audio clips are unique i.e., some clips might be identical since they are spliced from the same track/song.

Features:

Each feature has both its mean value and its standard deviation value stored in individual columns expect for label.

Root Mean Square Error (RMSE):

• RMSE: 📊

Spectral Features:

• Spectral Centroid: 🔊
• Spectral Bandwidth: 🎶
• Spectral Rolloff: 🌟

Zero Crossing Rate:

• Zero Crossing Rate: 🔀

Mel-Frequency Cepstral Coefficients (MFCC):

• MFCC1: 🎵
• MFCC2: 🔊
  (upto)
• MFCC39: 🌈
• MFCC40: 💫

Chroma Features:

• Chroma1: 🎹
  (upto)
• Chroma12: 🎼

Tonnetz Features:

• Tonnetz1: 🔊
  (upto)
• Tonnetz6: 🌈

Additional Features:

• Chroma CQT: 🔊
• Spectral Contrast: 🎶

Label:

• Label: 🏷️

Have fun and please make sure to upvote this because it took a me around 17 hours to build this dataset and I would greatly appreciate if you make some models with it and share it with others. Enjoy coding!