FMA: A Dataset for Music Analysis
By Michaël Defferrard, Kirell Benzi, Pierre Vandergheynst, Xavier Bresson. International Society for Music Information Retrieval Conference, 2016
637 citations recorded by the source corpus.
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 this https URL
Subfields: Computer Vision and Pattern Recognition, Music, Signal Processing
View Paper: https://www.semanticscholar.org/paper/9a82095be10926f0a52f8f9939deadfe39be2184
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Vision and Pattern Recognition
- Jesse Engel
- Antoine Caillon
- Xavier Serra
- Neil Zeghidour
- Adam Roberts
- Marco Tagliasacchi
- Zalán Borsos
- Qiuqiang Kong
In Music
- Xavier Serra
- Qiuqiang Kong
- Mark D. Plumbley
- Wenwu Wang
- Benjamin Elizalde
- Huaming Wang
- Aren Jansen
- Eduardo Fonseca
In Signal Processing
- Xavier Serra
- Neil Zeghidour
- Marco Tagliasacchi
- Jesse Engel
- Zalán Borsos
- Huaming Wang
- Antoine Caillon
- Adam Roberts
Citations in the corpus
- MusicLM: Generating Music From Text (2023, methodology): “We train SoundStream and w2v-BERT on the Free Music Archive (FMA) dataset (Defferrard et al., 2017), whereas the tokenizers and the autoregressive models for the semantic and acoustic modeling stages are trained on a dataset containing five million audio clips, amounting to 280k hours of music at 24…”
- AudioLDM 2: Learning Holistic Audio Generation With Self-Supervised Pretraining (2023, methodology): “The datasets used in this work include AudioSet (AS) [59], WavCaps [80], AudioCaps (AC) [27], VGGSound (VS) [81], Free Music Archive (FMA) [82], Million Song Dataset (MSD) [83], LJSpeech (LJS) [28], and GigaSpeech (GGS) [84].”
- The MTG-Jamendo Dataset for Automatic Music Tagging (2019, methodology): “In the context of auto-tagging, recent studies commonly use three datasets (Law et al., 2009; BertinMahieux et al., 2011; Defferrard et al., 2017) summarized in Table 1 and known for their limitations.”
- Pengi: An Audio Language Model for Audio Tasks (2023, background): “[63], MELD [48]; Music Analysis: NSynth [19], FMA [9]; Audio Captioning: AudioCaps [32], ClothoV2 [15]; Audio Question and Answering: ClothoAQA [39]; Auxiliary: WavText5K [12], SoundDescs [35], MACS [42], WavCaps [43], FreeSound [20] and FindSound 2 .”
- MAM: Masked Acoustic Modeling for End-to-End Speech-to-Text Translation (2020, positive): “We conducted MAM pre-training experiment on two corpora, Librispeech (human speech in English) (Panayotov et al., 2015) and FMA (music audios) (Defferrard et al., 2016), respectively.”
- Learning Discrete Structures for Graph Neural Networks (2019, methodology): “We also use FMA, a dataset where 140 audio features are extracted from 7,994 music tracks and where the problem is genre classification (Defferrard et al., 2017).”