There is a newer version of this record available.

Dataset Open Access

EMOPIA: A Multi-Modal Pop Piano Dataset For Emotion Recognition and Emotion-based Music Generation

Hung, Hsiao-Tzu; Ching, Joann; Doh, Seungheon; Kim, Nabin; Nam, Juhan; Yang, Yi-Hsuan


Citation Style Language JSON Export

{
  "publisher": "Zenodo", 
  "DOI": "10.5281/zenodo.5090631", 
  "title": "EMOPIA: A Multi-Modal Pop Piano Dataset For Emotion Recognition and Emotion-based Music Generation", 
  "issued": {
    "date-parts": [
      [
        2021, 
        7, 
        18
      ]
    ]
  }, 
  "abstract": "<p>EMOPIA (pronounced &lsquo;yee-m&ograve;-pi-uh&rsquo;) dataset is a shared multi-modal (audio and MIDI) database focusing on perceived emotion in&nbsp;<strong>pop piano music</strong>, to facilitate research on various tasks related to music emotion. The dataset contains&nbsp;<strong>1,087</strong>&nbsp;music clips from 387 songs and&nbsp;<strong>clip-level</strong>&nbsp;emotion labels annotated by four dedicated annotators.&nbsp;</p>\n\n<p>For more detailed information about the dataset, please refer to our paper:&nbsp;<a href=\"https://arxiv.org/abs/2108.01374\"><strong>EMOPIA: A Multi-Modal Pop Piano Dataset For Emotion Recognition and Emotion-based Music Generation</strong></a>.&nbsp;</p>\n\n<p><strong>File Description</strong></p>\n\n<ul>\n\t<li><em><strong>midis/</strong></em>:&nbsp;midi clips transcribed using GiantMIDI.\n\n\t<ul>\n\t\t<li>Filename `Q1_xxxxxxx_2.mp3`: Q1 means this clip belongs to Q1 on the V-A space; xxxxxxx is the song ID on YouTube, and the `2` means this clip is the 2nd clip taken from the full song.</li>\n\t</ul>\n\t</li>\n\t<li><em><strong>metadata/</strong></em>:&nbsp;metadata from YouTube. (Got when crawling)</li>\n\t<li>\n\t<p><em><strong>songs_lists/</strong></em>:&nbsp;YouTube URLs of songs.</p>\n\t</li>\n\t<li>\n\t<p><em><strong>tagging_lists/</strong></em>:&nbsp;raw tagging result for each sample.</p>\n\t</li>\n\t<li>\n\t<p><em><strong>label.csv</strong></em>: metadata that records filename, clip timestamps, and annotator.</p>\n\t</li>\n\t<li>\n\t<p><em><strong>metadata_by_song.csv</strong></em>: list all the clips by the song. Can be used to create the train/val/test splits to avoid the same song appear in both train and test.</p>\n\t</li>\n\t<li>\n\t<p><em><strong>scripts/prepare_split.ipynb:</strong></em> the script to create train/val/test splits and save them to csv files.</p>\n\t</li>\n</ul>\n\n<p>&nbsp;</p>\n\n<p><strong>Cite this dataset</strong></p>\n\n<pre><code>@inproceedings{{EMOPIA},\n         author = {Hung, Hsiao-Tzu and Ching, Joann and Doh, Seungheon and Kim, Nabin and Nam, Juhan and Yang, Yi-Hsuan},\n         title = {{MOPIA}: A Multi-Modal Pop Piano Dataset For Emotion Recognition and Emotion-based Music Generation},\n         booktitle = {Proc. Int. Society for Music Information Retrieval Conf.},\n         year = {2021}\n}</code></pre>", 
  "author": [
    {
      "family": "Hung, Hsiao-Tzu"
    }, 
    {
      "family": "Ching, Joann"
    }, 
    {
      "family": "Doh, Seungheon"
    }, 
    {
      "family": "Kim, Nabin"
    }, 
    {
      "family": "Nam, Juhan"
    }, 
    {
      "family": "Yang, Yi-Hsuan"
    }
  ], 
  "id": "5090631", 
  "version": "1.0", 
  "type": "dataset", 
  "event": "International Society for Music Information Retrieval Conference 2021 (ISMIR)"
}
953
493
views
downloads
All versions This version
Views 953708
Downloads 493171
Data volume 10.5 GB942.2 MB
Unique views 701595
Unique downloads 276161

Share

Cite as