DESED development dataset (real recordings)

sounds
Download Site Turpault2019

General information

Label Value Description
Name DESED development dataset (real recordings) Full dataset name
ID sounds/desed2019_dev_real Datalist id for external indexing
Abbreviation DESED-DEV-REAL Official dataset abbreviation, e.g. one used in the original paper
Year 2019 Dataset release year
Modalities Audio Video Data modalities included in the dataset
Collection name DESED Common name for all related datasets, used to group datasets coming from same source
Research domain SED Domestic Strong annotation Weak annotation Unlabelled Audio-visual Related domains, e.g., Scenes, Mobile devices, Audio-visual, Open set, Ambient noise, Unlabelled, Multiple sensors, SED, SELD, Tagging, FL, Strong annotation, Weak annotation, Unlabelled, Multi-annotator
Related datasets name
Download Download (23.4GB)
Companion site Site Link to the companion site for the dataset
Citation [Turpault2019] Sound event detection in domestic environments with weakly labeled data and soundscape synthesis

Audio

Label Value Description

Data

Data type Audio Possible values: Audio | Features
File format
File format type Constant Possible values: Constant | Variable
File format wav Possible value: wav | aiff | flac | mp3 | aac | ogg
Bit rate 16 Bit depth of audio, possible values: 8 | 16 | 24 | 32
Sampling rate (kHz) 44.1 kHz Sampling rate in kHz, possible values: 8 | 16 | 22.05 | 32 | 44.1 | 48
Channels
Material
Source Youtube Possible values: Original | Youtube | Freesound | Online | Crowdsourced | [Dataset name]

Content

Content type Freefield Possible values: Freefield | Synthetic | Isolated
Scene Variable Is the scene class constant for single recording, possible values: Constant | Variable
Event / Spatial location Unknown Possible values: Constant | Moving | Unknown

Recording

Setup Unknown Possible values: Near-field | Far-field | Mixed | Uncontrolled | Unknown
Spot type Unknown Possible values: Fixed | Moving | Unknown

Files

Count 19203 files Total number of files
Total duration (minutes) 2859.6 min Total duration of the dataset in minutes
File length Constant Characterization of the file lengths, possible values: Constant | Quasi-constant | Variable
File length (seconds) 10 sec Approximate length of files

Meta

Label Value Description
Types Event Tag List of meta data types provided for the data, possible values: Event, Tag, Scene, Caption, Geolocation, Spatial location, Annotator, Timestamp, Presence, Proximity, etc.

Event

Classes 10 Number of event classes
Classes False Possible values: True | False | Almost
Annotation
Type Strong Weak None Possible values: Strong | Weak | Location | None
Source Experts Possible values: Experts | Crowdsourced | Synthetic | Metadata | Automatic
Annotations per item 1 How many annotations there are available per item (possible multi-annotator setup)
Labelled amount (%) 25.2 % Percentage of all data, amount of data which is labelled
Strong annotations amount (%) 16.0 % Percentage of all data, amount of data which has strong annotations
Overlapping event instances Yes
Labeling
Hierarchical No
Instance
Count 6337 Count of all event instances in the dataset
Average instances per class 633.7 Average per class instance count

Cross-validation setup

Label Value Description
Provided Yes
Folds 1
Sets Train Val Set types provided in the split, possible values: Train | Test | Val | Dev | Eval

Baseline

Label Value Description
Download Download Link to baseline system source code
Citation [Turpault2019] Paper to cite for the baseline

Info

Label Value Description
Evaluation campaign DCASE2019 task4 Evaluation campaigns where the dataset was used.