The MTG-Jamendo Dataset for Automatic Music Tagging

295 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

High Fidelity Neural Audio Compression (2022, methodology): “For music, we rely on the Jamendo dataset (Bogdanov et al., 2019) for training and evaluation and we further evaluate our models on music using a proprietary music dataset.”
High-Fidelity Audio Compression with Improved RVQGAN (2023, methodology): “For music, we use the MUSDB dataset [31], and the Jamendo dataset [4].”
Qwen3-Omni Technical Report (2025, unknown): “We also employ multiple datasets including GTZAN (Tzanetakis & Cook, 2002), four subsets of MTG-Jamendo (MTG, Bogdanov et al. (2019)), and MagnaTagATune (Law et al., 2009) to evaluate the model’s capabilities across various music information retrieval tasks including genre identification, emotion…”
Evaluation of CNN-based Automatic Music Tagging Models (2020, methodology): “MTG-Jamendo Dataset [17] contains audio for 55,701 full songs and is built using music publicly available on the Jamendo 6 music platform under Creative Commons licenses.”
MERT: Acoustic Music Understanding Model with Large-Scale Self-supervised Training (2023, background): “However, this foundation model uses cumbersome hierarchical auto-regressive transformer decoders containing billions of parameters to model music audio, leading to significant inefficiency for conducting general music understanding tasks (e.g., it takes weeks to inference on datasets like MTG (Bogda”