GlobalMood

GlobalMood

A large-scale cross-cultural dataset for music emotion recognition, featuring 1,182 tracks from 5 countries with multilingual participant ratings.

View on GitHub

Associated publications

Lee, H., Çelen, E., Harrison, P. M. C., Anglada-Tort, M., van Rijn, P., Park, M., Schönwiesner, M., & Jacoby, N. (2025). GlobalMood: A cross-cultural benchmark for music emotion recognition. International Society for Music Information Retrieval Conference. https://doi.org/10.48550/arXiv.2505.09539

Overview

GlobalMood is a large-scale cross-cultural dataset for music emotion recognition (MER), featuring 1,182 music tracks from 5 countries (with plans to expand to over 20 countries), multilingual participant ratings in 5 languages (Arabic, Spanish, French, Korean, English), and mood descriptors freely elicited through an iterative chain process across participants.

Dataset Description

The dataset contains four main CSV files:

  • songmeta_GlobalMood.csv (70KB): Song metadata including YouTube video IDs, countries, artists, and titles
  • rawrating_GlobalMood.csv (30MB): Individual participant ratings for each song-mood pair
  • meanrating_GlobalMood.csv (6.3MB): Aggregated mean ratings across participants in each country
  • chains_GlobalMood.csv (5.9MB): Tags provided by participants in chains during the elicitation phase

Experimental Design

The experiment was implemented using the PsyNet framework. Participants rated how well each mood tag represents the mood expressed or conveyed by the music using a 5-point scale (1 = Not expressing at all, 5 = Extremely expressing).

Citation

If you use this dataset in your research, please cite:

Lee, H., Çelen, E., Harrison, P. M. C., Anglada-Tort, M., van Rijn, P., Park, M., Schönwiesner, M., & Jacoby, N. (2025). GlobalMood: A cross-cultural benchmark for music emotion recognition. Proceedings of the 26th International Society for Music Information Retrieval Conference (ISMIR).

License

This project is licensed under the MIT License.