The MusAV Dataset


MusAV is a new public benchmark dataset for comparative validation of arousal and valence (AV) regression models for audio-based music emotion recognition. We built MusAV by gathering comparative annotations of arousal and valence on pairs of music tracks, using track audio previews and metadata from the Spotify API. The resulting dataset contains 2,092 track previews covering 1,404 genres, with pairwise relative AV judgments by 20 annotators and various subsets of the ground truth based on different levels of annotation agreement.

This repository contains metadata, scripts, and instructions on how to download and use the dataset.


Read the ISMIR 2022 publication for more details about the dataset creation and annotation process. You can also check our ISMIR 2022 presentation materials (poster, video).

Data in data

We provide metadata for the entire annotation pool (a preselection of music tracks for annotation) we used to create the dataset, organized into triplets of tracks (for which we collect pairwise annotations) and split into chunks.

We have gathered human arousal/valence pairwise relative annotations for chunks 001-006, with three participating human annotators for each chunk. The dataset includes audio previews and Spotify API metadata for the annotated chunks. Please, contact us if you want to expand the dataset by annotating more chunks.

The data is organized in TSV and JSONL formats as follows:

Scripts in scripts

We provide scripts for evaluation of arousal/valence regression model predictions on the MusAV ground truth as well as scripts used for dataset creation and stats.

Statistics in stats

Citing the dataset

Please cite the following ISMIR 2022 publication when using the dataset:

Bogdanov, D., Lizarraga-Seijas, X., Alonso-Jiménez, P., & Serra X. (2022). MusAV: A dataset of relative arousal-valence annotations for validation of audio models. International Society for Music Information Retrieval Conference (ISMIR 2022).

@conference {bogdanov2019mtg,
    author = "Bogdanov, Dmitry and Lizarraga-Seijas, Xavier and Alonso-Jiménez, Pablo and Serra, Xavier",
    title = "MusAV: A dataset of relative arousal-valence annotations for validation of audio models",
    booktitle = "International Society for Music Information Retrieval Conference (ISMIR 2022)",
    year = "2022",
    address = "Bengaluru, India",
    url = ""



This research was carried out under the project Musical AI - PID2019-111403GB-I00/AEI/10.13039/501100011033, funded by the Spanish Ministerio de Ciencia e Innovación and the Agencia Estatal de Investigación.

We thank all the annotators who participated in the creation of the dataset.