Metal Album Art by Subgenre

Data Scraping: We leveraged the extensive library of music metadata provided by the Spotify API to scrape data efficiently.

Band Selection: We actively searched for bands by genre, ensuring we represented various metal subgenres comprehensively.

Image Retrieval: We used the URLs provided by the Spotify API to extract and save album art images for each band.

Subgenre Classification

To maintain simplicity and clarity, the dataset focuses on high-level subgenre classifications. We aim to avoid delving into intricate influences or hybrid subgenres. For instance, we classify Blind Guardian under Power Metal, without considering their Thrash Metal influences. Similarly, we categorize Fleshgod Apocalypse under Death Metal, despite their Symphonic Death Metal elements.

Fleshgod Apocalypse is categorized under Death Metal, despite their Symphonic Death Metal elements.

Data Cleaning and Exclusions

To maintain the dataset’s accuracy and relevance, several steps were taken:

Incorrect Labeling: Bands labeled incorrectly by the Spotify API were excluded.

Band Verification: Only bands clearly identifiable as metal were included, avoiding confusion with similarly named artists from different genres.

Content and Usage

This dataset serves multiple purposes:

Educational Tool: It is perfect for learning and practicing data scraping, image processing, and classification using deep learning techniques. Furthermore, it provides an excellent opportunity for hands-on experience in these areas.

Research: The dataset offers a valuable resource for studying metal music subgenres, album art styles, and their evolution. Moreover, it supports academic studies and papers in these fields.

Project Development: Users can utilize this dataset to develop projects related to music recommendation systems, genre classification models, and more. Additionally, it can inspire innovative applications in music technology.

Acknowledgments

Special thanks to the Spotify API for providing access to their extensive music database, and to the Fast.ai community for their support and resources.

Additional Notes

This dataset serves as a starting point for further exploration and refinement. We encourage users to build upon this work by enhancing the accuracy and depth of the subgenre classifications. Additionally, the dataset can be expanded to include more bands and subgenres, thereby providing a richer resource for both the metal music community and machine learning enthusiasts.