{
  "fetched_at": "2026-10-20T13:08:50Z",
  "source_url": "https://www.kaggle.com/competitions/dcase2019-task1a-leaderboard/data",
  "pages": [
    {
      "name": "rules",
      "content": "# External data\nUse of external data **is allowed**, except [TUT Acoustic Scenes 2016](https://zenodo.org/record/45739), [TUT Acoustic Scenes 2017](https://zenodo.org/record/400515), [TUT Urban Acoustic Scenes 2018](https://zenodo.org/record/1228142) and leaderboard datasets (DCASE2018 and DCASE2019).\n\n# Data manipulation\n\nManipulation of provided training and development data **is allowed** . The development dataset (training dataset) can be augmented (e.g. by mixing data sampled from a pdf or using techniques such as pitch shifting or time stretching). \n\n# Test data\n\nParticipants are **not allowed** to make subjective judgments of the test data, nor to annotate it. The test data cannot be used to train the submitted system; the use of statistics about the test data in the decision making is also forbidden. Classification decision must be done independently for each test sample. One cannot use information from other test items when predicting the scene label. \n\n# Dataset license\n\nAdditionally, you agree that the use of provided data is limited to this competition and bound by the LICENSE (see \"Data\" section of the competition). Essentially, the LICENSE prohibits any commercial use of the dataset.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "data-description",
      "content": "The data for the competition consists of **training dataset**, 10-second audio segments released together with the scene class labels, and **test dataset** which is released without the scene labels. Your task is to predict the scene labels for the audio segments in the test dataset.\n\n# Datasets\n\n## Training dataset\n\nThe development dataset provided with DCASE2019 Challenge task 1A is used as training dataset. There are three methods for obtaining the training dataset.\n\n1) Dataset can be downloaded directly\n\n***\n\n### [TAU Urban Acoustic Scenes 2019, Development dataset  (35.6 GB) ](https://zenodo.org/record/2589280)\n\n[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.2589280.svg)](https://doi.org/10.5281/zenodo.2589280)\n\n***\n\n2) Dataset can be downloaded using the provided download script:\n\n`python task1a_dataset_downloader.py -d train`\n\n3) If you are using official [DCASE2019 Challenge Task 1 baseline system](https://github.com/toni-heittola/dcase2019_task1_baseline), you can use it to download the dataset too:\n\n`python task1a.py -m dev --download_dataset datasets/`\n\n### Organization of the dataset\n\nThe dataset package contains readme file, which explains how the dataset is organized. Reference data for the provided audio segments can be found in CSV-formated file `meta.csv`:\n\n<pre>filename   scene_label identifier  source_label\naudio/airport-barcelona-0-0-a.wav   airport barcelona-0 a\naudio/airport-barcelona-0-1-a.wav   airport barcelona-0 a\n....\naudio/bus-barcelona-17-706-a.wav    bus barcelona-17    a\naudio/bus-helsinki-18-707-a.wav bus helsinki-18 a\naudio/bus-helsinki-18-708-a.wav bus helsinki-18 a\n...\naudio/tram-vienna-285-8637-a.wav    tram    vienna-285  a\naudio/tram-vienna-285-8638-a.wav    tram    vienna-285  a\naudio/tram-vienna-285-8639-a.wav    tram    vienna-285  a\n</pre>\n\nNaming convention for audio segments is <pre>file naming convention: [scene label]-[city]-[location id]-[segment id]-[device id].wav</pre>\n\n`source_label` field in the meta data or `[device id]` field in the audio segment filename is used to denote the used recording device. In this competition, all material is captured with the same device (`a`).\n\nThe dataset is provided with single fold cross-validation set (training / testing) to ease the usage of the development dataset. Participants can do their own cross-validation setup too. File lists assigning segments to the subsets are provided under directory `evaluation_setup`. \n\n## Testing dataset\n\nSimilarly to the development dataset, there are three methods for obtaining the dataset.\n\n1) Downloading it directly:\n\n***\n\n### [TAU Urban Acoustic Scenes 2019, Leaderboard dataset  (3.0 GB) ](https://zenodo.org/record/2672993)\n\n[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.2672993.svg)](https://doi.org/10.5281/zenodo.2672993)\n\n***\n\n2) Dataset can be downloaded using the provided download script:\n\n`python task1a_dataset_downloader.py -d test`\n\n3) If you are using official [DCASE2019 Challenge Task 1 baseline system](https://github.com/toni-heittola/dcase2019_task1_baseline) you can use it to download the dataset too:\n\n`python task1a.py -m leaderboard --download_dataset datasets/`\n\n## Dataset license\n\nLicense permits free academic usage. Any commercial use is strictly prohibited. For commercial use, contact dataset authors.\n\n<pre>\nCopyright (c) 2019 Tampere University and its licensors\nAll rights reserved.\nPermission is hereby granted, without written agreement and without license or royalty\nfees, to use and copy the TAU Urban Acoustic Scenes 2019 (“Work”) described in this document\nand composed of audio and metadata. This grant is only for experimental and non-commercial\npurposes, provided that the copyright notice in its entirety appear in all copies of this Work,\nand the original source of this Work, (Audio Research Group at Tampere University of Technology),\nis acknowledged in any publication that reports research using this Work.\nAny commercial use of the Work or any part thereof is strictly prohibited.\nCommercial use include, but is not limited to:\n- selling or reproducing the Work\n- selling or distributing the results or content achieved by use of the Work\n- providing services by using the Work.\n\nIN NO EVENT SHALL TAMPERE UNIVERSITY OR ITS LICENSORS BE LIABLE TO ANY PARTY\nFOR DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE\nOF THIS WORK AND ITS DOCUMENTATION, EVEN IF TAMPERE UNIVERSITY OR ITS\nLICENSORS HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\n\nTAMPERE UNIVERSITY AND ALL ITS LICENSORS SPECIFICALLY DISCLAIMS ANY\nWARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND\nFITNESS FOR A PARTICULAR PURPOSE. THE WORK PROVIDED HEREUNDER IS ON AN \"AS IS\" BASIS, AND\nTHE TAMPERE UNIVERSITY HAS NO OBLIGATION TO PROVIDE MAINTENANCE, SUPPORT,\nUPDATES, ENHANCEMENTS, OR MODIFICATIONS.\n</pre>",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "![enter image description here][1]\n\nThis is a parallel competition to [**DCASE2019 challenge task 1A**](http://dcase.community/challenge2019/task-acoustic-scene-classification#subtask-a), intended to serve as public leaderboard for participants during the DCASE challenge. The competition is open for everybody and there is no requirement to submit to DCASE challenge. \n\nThis subtask is concerned with the basic problem of acoustic scene classification, in which all data (development and evaluation) are recorded with the same device, in this case device A, and contains only data from the 10 known acoustic scene classes. The subtask uses TAU Urban Acoustic Scenes 2019 dataset. \n\nThe training material of this competition is the development dataset released for Task1A, and testing material consist of **similar material** to the official evaluation dataset in the DCASE challenge. The testing material amount in this competition is considerably lower than the official evaluation material in the DCASE challenge. \n\nThere are three parallel competitions to DCASE challenge task 1 open in the Kaggle, one for each subtask:\n\n\n- **[Task1A Leaderboard, Acoustic Scene Classification](https://www.kaggle.com/c/dcase2019-task1a-leaderboard)**, single recording device used in the data collection, no external data allowed. \n\n- **[Task1B Leaderboard, Acoustic Scene Classification with mismatched recording devices](https://www.kaggle.com/c/dcase2019-task1b-leaderboard)**, multiple recording devices used in the data collection, no external data allowed.\n\n- **[Task1C Leaderboard, Open set Acoustic Scene Classification ](https://www.kaggle.com/c/dcase2019-task1c-leaderboard)**, classification on data that includes classes not encountered in the training data.\n\n***\n## **Note:** Official DCASE submission is **not done through Kaggle**. See [DCASE website](http://dcase.community/challenge2019/) for full information on how to submit to the challenge.\n***\n\n### Organizers\n\n- [Annamaria Mesaros](http://www.cs.tut.fi/~mesaros/), Assistant Professor, Tampere University, Finland\n- [Toni Heittola](http://www.cs.tut.fi/~heittolt/), Researcher, Tampere University, Finland\n- [Tuomas Virtanen](http://www.cs.tut.fi/~tuomasv/), Professor, Tampere University, Finland\n\n# Acoustic scene classification task\n\nThe goal of acoustic scene classification is to classify a test recording into one of the provided predefined classes that characterizes the environment in which it was recorded — for example \"park\", \"pedestrian street\", \"metro station\". \n\n![enter image description here][2]\n\n# Audio dataset\n\nThe dataset for this task is the **TAU Urban Acoustic Scenes 2019** dataset, consisting of recordings from various acoustic scenes. This dataset extends the TUT Urban Acoustic Scenes 2018 dataset with other 6 cities to a total of 12 large European cities. For each scene class, recordings were done in different locations; for each recording location there are 5-6 minutes of audio. The original recordings were split into segments with a length of 10 seconds that are provided in individual files. Available information about the recordings include the following: acoustic scene class, city, and recording location.\n\nAcoustic scenes (10):\n\n- Airport - `airport`\n- Indoor shopping mall - `shopping_mall`\n- Metro station - `metro_station`\n- Pedestrian street - `street_pedestrian`\n- Public square - `public_square`\n- Street with medium level of traffic - `street_traffic`\n- Travelling by a tram - `tram`\n- Travelling by a bus - `bus`\n- Travelling by an underground metro - `metro`\n- Urban park - `park`\n\nData was recorded in the following cities:\n\n- Amsterdam \n- Barcelona \n- Helsinki\n- Lisbon  \n- London\n- Lyon\n- Madrid\n- Milan\n- Prague\n- Paris \n- Stockholm \n- Vienna\n\nThe dataset was collected by Tampere University of Technology between 05/2018 - 11/2018. The data collection received funding from the European Research Council, grant agreement 637422 EVERYSOUND.\n\n![enter image description here][3]\n\n\n**TAU Urban Acoustic Scenes 2019, Development dataset** contains only material recorded with device A, having 1440 segments for each acoustic scene (240 minutes of audio). The dataset contains in total 14400 segments, i.e. 40 hours of audio.  \n\n***\n\n### [TAU Urban Acoustic Scenes 2019, Development dataset  (35.6 GB) ](https://doi.org/10.5281/zenodo.2589280)\n\n[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.2589280.svg)](https://doi.org/10.5281/zenodo.2589280)\n\n***\n\n**TAU Urban Acoustic Scenes 2019, Leaderboard dataset** contains only material recorded with device A, and it is used in this competition as test dataset. This dataset is used only in this competition, not in the actual DCASE2019 Challenge.\n\n***\n\n### [TAU Urban Acoustic Scenes 2019, Leaderboard dataset  (3.0 GB) ](https://zenodo.org/record/2672993)\n\n[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.2672993.svg)](https://doi.org/10.5281/zenodo.2672993)\n\n***\n\n## Recording procedure\n\nRecordings were made using four devices that captured audio simultaneously.  \n\nThe main recording device consists in [Soundman OKM II Klassik/studio A3](http://www.soundman.de/en/products/), electret binaural microphone and a [Zoom F8](https://www.zoom.co.jp/products/handy-recorder/zoom-f8-multitrack-field-recorder) audio recorder using 48kHz sampling rate and 24 bit resolution. The microphones are specifically made to look like headphones, being worn in the ears. As an effect of this, the recorded audio is very similar to the sound that reaches the human auditory system of the person wearing the equipment. This equipment is further referred to as device A.\n\nThe other devices are commonly available customer devices: device B is a Samsung Galaxy S7, device C is IPhone SE, and device D is a GoPro Hero5 Session. All simultaneous recordings are time synchronized. \n\n## Baseline system\n\nThe baseline system provides a simple entry-level approach that gives reasonable results in the subtasks of Task 1. The baseline system is built on [dcase_util toolbox](https://github.com/DCASE-REPO/dcase_util).\n\n***\n\n### [DCASE2019 Task 1 Baseline code repository](https://github.com/toni-heittola/dcase2019_task1_baseline)\n\n***\n\n### System description\n\nThe baseline system implements a convolutional neural network (CNN) based approach, where log mel-band energies are first extracted for each 10-second signal, and a network consisting of two CNN layers and one fully connected layer is trained to assign scene labels to the audio signals. \n\nMore detailed description and system accuracy on development dataset can be found from readme file in the [code repository](https://github.com/toni-heittola/dcase2019_task1_baseline/blob/master/README.md).\n\n\n  [1]: http://dcase.community/images/logos/dcase/dcase2019_challenge.png\n  [2]: http://d33wubrfki0l68.cloudfront.net/15556a316bea611e8f73980845a7ef41dd162de9/05afc/images/tasks/challenge2019/task1_acoustic_scene_classification.png\n  [3]: http://d33wubrfki0l68.cloudfront.net/750a7452e2c66bfb113a1b011ecc888f03245c99/0e821/images/sponsors/erc.jpg",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "The participants will be ranked based on the accuracy of their predictions. The evaluation metric for this competition is the prediction accuracy, the proportion of 10-second test recordings for which your algorithm makes the correct prediction.\n\n## Submission Format\n\nSubmission files should be in csv format and contain two columns: Id and Prediction. The first column of the csv file is the test sample ID (0,1,2,3,...,1199) and the second column is the class name predicted for that sample by your algorithm.\n\nThe file should contain a header and have the following format:\n\n<pre>Id,Scene_label<br>0,metro_station<br>1,bus<br>2,metro<br>...<br>1199,park</pre>\n\n\n***\n## **Note:** The submission format is different to the one used in DCASE challenge due to requirements of the Kaggle system.\n***",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}