Datasets


Environmental sounds

Introduction

Audio data collection and manual data annotation both are tedious processes, and lack of proper development dataset limits fast development in the environmental audio research.

Currently, there are only handful of large datasets available and some of them might be hard to find (e.g. they are used in neighboring research fields). This page tries to maintain a list of datasets suitable for environmental audio research. In addition to the freely available dataset, also proprietary and commercial datasets are listed here for completeness.

In addition to the datasets, also some of the on-line sound services are listed at the end of the page. These services can be used to form new datasets for special research needs.

If you know datasets which should be in this page or you have knowledge (statistics) about some of the commercial dataset, please let me know.

Environmental audio

The datasets are divided into two tables: Sound events table contains datasets suitable for research in the field of automatic sound event detection and automatic sound tagging. Acoustic scenes table contains datasets suitable for research involving the audio-based context recognition and acoustic scene classification.

Sound events

Provider Name Audio
type
Context
selection
Annotation
type
Context
count
Event
classes
Event
count
Instances
per class
Audio
files
Mins Context
classes
Licence Link Pub
Dares G1 Live Mixed Events 28 761 3214 4.2 123 123 Free http://www.daresounds.org/ http://ieeexplore.ieee.org/document/6637761;Mesaros2013a, http://ieeexplore.ieee.org/document/6616142;Mesaros2013b
Dares Amstel Live Single Events 1 13 1002 77.1 40 54 train station Free http://www.daresounds.org/
Sound Ideas BBC Sound Effects Library Isolated Mixed Description 0 0 1655 1655 0 Commercial http://www.sound-ideas.com/sound-effects/bbc-1-40-cds-sound-effects-library.html http://ieeexplore.ieee.org/document/6269853;Kim2012a, http://journals.cambridge.org/action/displayFulltext?type=6&fid;=8779081&jid;=SIP&volumeId;=1&issueId;=-1&aid;=8779080&fulltextType;=RA&fileId;=S2048770312000078;Kim2012b, http://sail.usc.edu/~malandra/files/papers/interspeech2013.pdf;Malandrakis2013
TUT CASA 2009 Live Mixed Events 10 208 10326 49.6 103 1133 basketball game, beach, hallway, inside a bus, inside a car, office, restaurant, stadium with track and field event, street, supermarket Proprietary http://ieeexplore.ieee.org/document/6639360;Heittola2013a, http://asmp.eurasipjournals.com/content/2013/1/1/abstract;Heittola2013b, http://www.cs.tut.fi/~heittolt/pubs/chime2011_heittola.pdf;Heittola2011, http://www.cs.tut.fi/~mesaros/pubs/plsa.pdf;Mesaros2011, http://www.cs.tut.fi/~mesaros/pubs/acoustic_event_detection_1406.pdf;Mesaros2010, http://www.cs.tut.fi/~heittolt/pubs/eusipco2010_heittola.pdf;Heittola2010
TUT CASA 2010 Live Mixed Events 16 289 4173 14.4 160 535 Proprietary
ELRA CHIL 2007 Evaluation Package Live Single Events 1 0 0 0 0 meeting room Commercial http://catalog.elra.info/product_info.php?products_id=1092 http://dl.acm.org/citation.cfm?id=1840231;Zhuang2010
IEEE AASP Challenge 2013 Event isolated Isolated Single Tags 1 16 639 39.9 320 19 office Free http://c4dm.eecs.qmul.ac.uk/sceneseventschallenge/description.html http://c4dm.eecs.qmul.ac.uk/sceneseventschallenge/;Results
IEEE AASP Challenge 2013 Event live Live Single Events 1 16 205 12.8 3 5 office Free http://c4dm.eecs.qmul.ac.uk/sceneseventschallenge/description.html http://c4dm.eecs.qmul.ac.uk/sceneseventschallenge/;Results
IEEE AASP Challenge 2013 Event synthetic Live Single Events 1 15 310 20.7 9 14 office Free http://c4dm.eecs.qmul.ac.uk/sceneseventschallenge/description.html http://c4dm.eecs.qmul.ac.uk/sceneseventschallenge/;Results
NII-SRC RWCP Sound Scene Database Isolated Mixed Tags 0 14 9722 694.4 9722 0 Research and development use only http://www.openslr.org/13/ http://ieeexplore.ieee.org/document/1035570;Nishiura2002, http://link.springer.com/article/10.1007%2Fs00779-005-0045-4;Smith2006, http://ieeexplore.ieee.org/document/6637759;Dennis2013
QMUL Freefield1010 Live Mixed Tags 0 0 7690 7690 1282 Creative Commons http://c4dm.eecs.qmul.ac.uk/rdr/handle/123456789/35 http://arxiv.org/abs/1309.5275;Stowell2014
INRIA NAR Isolated Mixed Tags 4 41 42 1.0 852 8 Creative Commons https://team.inria.fr/perception/nard/ https://hal.inria.fr/hal-00768767/en;Janvier2014,https://hal.inria.fr/hal-00952092/en;Janvier2012
NYU UrbanSound Live Single Events 1 10 3075 307.5 1302 1620 street Creative Commons http://serv.cusp.nyu.edu/projects/urbansounddataset/urbansound.html http://serv.cusp.nyu.edu/projects/urbansounddataset/salamon_urbansound_acmmm14.pdf;Salamon2014
NYU UrbanSound8K Live Street Tags 1 10 8732 873.2 8732 525 street Creative Commons http://serv.cusp.nyu.edu/projects/urbansounddataset/urbansound8k.html http://serv.cusp.nyu.edu/projects/urbansounddataset/salamon_urbansound_acmmm14.pdf;Salamon2014
ELRA FBK-Irst database of isolated meeting-room acoustic events Isolated Single Tags 1 16 576 36.0 288 63 meeting room Free http://catalog.elra.info/product_info.php?products_id=1093 http://www.ee.columbia.edu/~dpwe/pubs/CottonE11-spectrotemporal.pdf;Cotton2011
ELRA UPC-TALP database of isolated meeting-room acoustic events Live Single Events 1 14 1026 73.3 3 0 meeting room Commercial http://catalog.elra.info/product_info.php?products_id=1053
TU Dortmund Acoustic event dataset Isolated Single Events 1 12 235 19.6 23 34 meeting room Free http://patrec.cs.tu-dortmund.de/files/icassp2014aed_dortmund.zip http://patrec.cs.tu-dortmund.de/pubs/papers/Plinge2014-BOF.pdf;Plinge2014
CICESE Sound Events Isolated Single Tags 1 20 1367 68.3 1367 92 Free http://sound.natix.org/databases/allSounds.zip http://www.sciencedirect.com/science/article/pii/S0167865515002925;Beltran2015
MIVIA Audio Events Data Set for Surveillance Applications Live Single Events 1 3 6000 2000.0 760 2279 Free http://mivia.unisa.it/datasets/audio-analysis/mivia-audio-events/ http://www.sciencedirect.com/science/article/pii/S0167865515001981;Foggia2015
MIVIA Audio Events Data Set for Road Surveillance Applications Live Single Events 1 2 400 200.0 57 60 Free http://mivia.unisa.it/datasets/audio-analysis/mivia-road-audio-events-data-set/ http://ieeexplore.ieee.org/document/6918643;Foggia2014
Freiburg Freiburg-106, Audio Data Set for Human Activity Recognition Live Single Tags 1 24 1524 63.5 1524 54 kitchen Free http://www.csc.kth.se/~jastork/pages/datasets.html http://ieeexplore.ieee.org/document/6343802;Stork2012, http://www.eurasip.org/Proceedings/Eusipco/Eusipco2015/papers/1570103447.pdf;Phan2015
ESC ESC-50 Live Mixed Tags 0 50 2000 40.0 2000 166 Free https://github.com/karoldvl/ESC-50 http://karol.piczak.com/papers/Piczak2015-ESC-Dataset.pdf;Piczak2015, http://karol.piczak.com/papers/Piczak2015-ESC-ConvNet.pdf;Piczak2015b
ESC ESC-10 Live Mixed Tags 0 10 400 40.0 400 33 Free https://github.com/karoldvl/ESC-10 http://karol.piczak.com/papers/Piczak2015-ESC-Dataset.pdf;Piczak2015, http://karol.piczak.com/papers/Piczak2015-ESC-ConvNet.pdf;Piczak2015b
ESC Dataset for Environmental Sound Classification Live Mixed Tags 0 50 2000 40.0 250000 20833 Free https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/YDEPUT http://karol.piczak.com/papers/Piczak2015-ESC-Dataset.pdf;Piczak2015a, http://karol.piczak.com/papers/Piczak2015-ESC-ConvNet.pdf;Piczak2015b
TUT TUT Sound events 2016, Development dataset Live Mixed Events 2 18 954 53.0 22 78 home, residential area Free https://zenodo.org/record/45759
TU Dortmund Multi-channel acoustic event dataset Isolated Single Events 1 20 437 21.9 37 90 conference room Free http://patrec.cs.tu-dortmund.de/files/datasets/dcase2016_multichannel-aed_dortmund.7z http://patrec.cs.tu-dortmund.de/pubs/papers/Kuerby2016-BAE;Kurby2016
TUT TUT-SED Synthetic 2016 Synthetic Single Events 1 16 100 566 synthetic Free http://www.cs.tut.fi/sgn/arg/taslp2017-crnn-sed/tut-sed-synthetic-2016
TUT TUT Sound events 2017, Development dataset Live Mixed Events 2 6 729 121.5 24 92 street Free https://zenodo.org/record/400516
TUT TUT Rare sound events 2017, Development dataset Synthetic Single Events 1 3 1281 625 synthetic Free http://www.cs.tut.fi/sgn/arg/dcase2017/challenge/download#task-2---detection-of-rare-sound-events
Google AudioSet Live Mixed Tags 632 2084320 3297.9 2084320 347386 Free https://research.google.com/audioset/ https://research.google.com/pubs/pub45857.html;Gemmeke2017


Instructions to use the table

Sort the table by clicking headers. Select more fields to be shown from -menu. The table can be filtered by selection condition from the select boxes. Filtering can be cleared with -button. Numerical data is also shown in bar-chart. The chart can be hiden and shown from Chart-button. Pagination can be enabled and disabled with -button.


Notation

Audio type is denoted as follows:

  • Isolated, only one sound event is active per sample (no overlapping sounds).
  • Live, real recordings where overlapping sounds may be presents, start and end times annotated.

Annotation type is denoted as follows:

  • Sound Events, timestamps (onset and offset times) are indicated
  • Tags, sample-wide (usually one word or short) textual label
  • Description, sample-wide textual description of sounds / sound scene

Acoustic scenes

Provider Name Context
count
Audio
files
Instances
per class
Mins Context
classes
Licence Link Pub
UEA Noise DB / Series 1 10 10 1.0 40 Free http://lemur.cmp.uea.ac.uk/Research/noise_db/ http://link.springer.com/article/10.1007%2Fs00779-005-0045-4;Smith2006
UEA Noise DB / Series 2 12 35 2.9 175 Free http://lemur.cmp.uea.ac.uk/Research/noise_db/ http://link.springer.com/article/10.1007%2Fs00779-005-0045-4;Smith2006
IEEE AASP Challenge 2013 Scene 10 100 10.0 50 Free http://c4dm.eecs.qmul.ac.uk/sceneseventschallenge/description.html http://c4dm.eecs.qmul.ac.uk/sceneseventschallenge/;Results
TUT CASA2009 10 103 10.3 1133 Proprietary http://www.cs.tut.fi/~heittolt/pubs/eusipco2010_heittola.pdf;Heittola2010
TUT CASA2010 13 160 12.3 533 Proprietary
TUT CASR 27 225 8.3 0 Proprietary http://ieeexplore.ieee.org/document/1561288/;Eronen2006
Dares G1 28 123 4.4 123 Free http://www.daresounds.org/
Inria DEMAND 0 18 0 Creative Commons http://parole.loria.fr/DEMAND/
LITIS Rouen audio scene dataset 19 3026 159.3 1513 Free https://sites.google.com/site/alainrakotomamonjy/home/audio-scene http://ieeexplore.ieee.org/document/6971128;Rakotomamonjy2015
TUT TUT Acoustic scenes 2016 Development dataset 15 1170 78.0 585 Free https://zenodo.org/record/45739 http://www.cs.tut.fi/~mesaros/pubs/mesaros_eusipco2016-dcase.pdf;Mesaros2016
TUT TUT Acoustic scenes 2017 Development dataset 15 4680 312.0 780 Free https://zenodo.org/record/400515
AucoDefr07 AucoDefr07 4 16 63.0 252 Free https://archive.org/details/defreville-Aucouturier_urbanDb https://hal.archives-ouvertes.fr/hal-01082501v2;Lagrange2015

Online services

Isolated sounds

Geotagged recordings

Related datasets

  • Speech Datasets, ISCA Special Interest Group on Robust Speech Recognition

Tools

Annotation tools

Software

  • Audacity, audio software with basic annotation capabilities. Use label tracks for the annotations, see more info here.
  • ELAN, a linguistic annotation tool to create the textual annotations for audio and video files

Prototypes

  • Soundscape, a tool for soundscape annotation
  • I-SED, an interactive sound event detector

Audio management tools

  • Pumilio, a Web-Based Management System for Ecological Recordings

Audio augmentation tools