SML – Music genre

Music genre identification

Usefull resources

• A notebook showing how to preprocess musical signal in Python
• A lecture with a review on feature choice/preprocessing for music…

Datasets

• GTZAN Genre Collection
  Dataset used for the well known paper in genre classification ” Musical genre classification of audio signals” by G. Tzanetakis and P. Cook in IEEE Transactions on Audio and Speech Processing 2002.
  The dataset consists of 10 genres : Blues, Classical, Country, Disco, Hiphop, Jazz, Metal, Pop, Reggae, Rock. Each genre contains 100 songs. Tracks are 30 seconds long and are all 22050Hz Mono 16-bit audio files in .wav format.
  Useful links:
  • The gtzan dataset: Its contents, its faults, their effects on evaluation, and its future use Bob L SturmarXiv preprint arXiv:1306.1461, 2013. (reference [48] cited by in https://arxiv.org/abs/1703.09179 : “… The Gtzan genre dataset has been extremely popular, although some flaws have been found [48] …”.)
• FMA: A Dataset For Music Analysis
  Dataset: fma_small.zip: 8,000 tracks of 30s, 8 balanced genres (GTZAN-like) (7.2 GiB)
  The dataset is a dump of the Free Music Archive (FMA), an interactive library of high-quality, legal audio downloads.
  Note by authors: This is a pre-publication release. As such, this repository as well as the paper and data are subject to change. Stay tuned!
  Used by: WWW 2018 Challenge: Learning to Recognize Musical Genre from Audio on the Web By EPFL
  Reference:
  • FMA: A Dataset For Music Analysis, M. Defferrard, K. Benzi, P. Vandergheynst, X. Bresson
    Paper abstract: We introduce the Free Music Archive (FMA), an open and easily accessible dataset suitable for evaluating several tasks in MIR, a field concerned with browsing, searching, and organizing large music collections. The community’s growing interest in feature and end-to-end learning is however restrained by the limited availability of large audio datasets. The FMA aims to overcome this hurdle by providing 917 GiB and 343 days of Creative Commons-licensed audio from 106,574 tracks from 16,341 artists and 14,854 albums, arranged in a hierarchical taxonomy of 161 genres. It provides full-length and high-quality audio, pre-computed features, together with track- and user-level metadata, tags, and free-form text such as biographies. We here describe the dataset and how it was created, propose a train/validation/test split and three subsets, discuss some suitable MIR tasks, and evaluate some baselines for genre recognition. Code, data, and usage examples are available at https://github.com/mdeff/fma.

• Audio Set
  Especially Music genre, cited by Music Genre Classification using Machine Learning Techniques which used it on 7 labels:

1. 1. Pop Music (8100)
   2. Rock Music (7990)
   3. Hip Hop Music (6958)
   4. Techno (6885)
   5. Rhythm Blues (4247)
   6. Vocal (3363)
   7. Reggae Music (2997)

• Music Information Retrieval Evaluation eXchange (MIREX 2019 and previous….)
  Many tasks at : https://www.music-ir.org/mirex/wiki/2019:Audio_Classification_(Train/Test)_Tasks
  (22.05 kHz mono wav clips)
  • Audio Classical Composer Identification
  • Audio US Pop Music Genre Classification
  • Audio Latin Music Genre Classification
  • Audio Mood Classification
  • Audio K-POP Mood Classification
  • Audio K-POP Genre Classification