TUT Acoustic scenes 2016, Evaluation dataset

Audio Research Group / Tampere University of Technology

Authors
- Toni Heittola (toni.heittola@tut.fi, http://www.cs.tut.fi/~heittolt/)
- Annamaria Mesaros (annamaria.mesaros@tut.fi, http://www.cs.tut.fi/~mesaros/)
- Tuomas Virtanen (tuomas.virtanen@tut.fi, http://www.cs.tut.fi/~tuomasv/)

Recording and annotation
- Eemi Fagerlund
- Aku Hiltunen

Table of Contents

  1. Dataset
  2. Usage
  3. Changelog
  4. License

1. Dataset

TUT Acoustic Scenes evaluation dataset consists of 30-seconds audio segments from 15 acoustic scenes:

Each acoustic scene has 26 segments totaling 13 minutes of audio.

The dataset was collected in Finland by Tampere University of Technology between 06/2015 - 01/2016. The data collection has received funding from the European Research Council under the ERC Grant Agreement 637422 EVERYSOUND.

ERC

Preparation of the dataset

For all acoustic scenes, the recordings were captured each in a different location: different streets, different parks, different homes. The equipment used for recording consists of a binaural Soundman OKM II Klassik/studio A3 electret in-ear microphone and a Roland Edirol R-09 wave recorder using 44.1 kHz sampling rate and 24 bit resolution.

Postprocessing of the recorded data involves aspects related to privacy of recorded individuals, and possible errors in the recording process. For audio material recorded in private places, written consent was obtained from all people involved. Material recorded in public places does not require such consent, but was screened for content, and privacy infringing segments were eliminated. Temporary microphone failure and radio signal interferences from mobile phones were annotated and these annotations are provided.

Audio material was cut into segments of 30 seconds length.

File structure

dataset root
│   README.md               this file, markdown-format
│   README.html             this file, html-format
│   EULA.pdf                End user license agreement
│   meta.txt                meta data, csv-format, [audio file (string)][tab][scene label (string)]
│
└───audio                   1170 audio segments, 24-bit 44.1kHz
│   │   1.wav       name format [original_recording_identifier]_[start sec (int)]_[end sec (int)].wav
│   │   2.wav
│   │   ...
│
└───evaluation_setup        cross-validation setup, 4 folds
    │   test.txt            testing file list, csv-format, [audio file (string)]
    │   evaluate.txt        evaluation file list, csv-format, [audio file (string)][tab][scene label (string)][tab][original audio file (string)]

2. Usage

The partitioning of the data was done based on the location of the original recordings. All segments obtained from the same original recording were included into a single subset - either development dataset or evaluation dataset. For each acoustic scene, 78 segments were included in the development dataset and 26 segments were kept for evaluation. Development dataset is provided separately.

Evaluation setup

Testing

evaluation setup\test.txt
testing file list (in csv-format)

Format:

[audio file (string)]

Evaluating

evaluation setup\evaluate.txt
evaluation file list (in csv-format), test.txt but added with ground truth information and original audio file name. These two files are provided separately to prevent contamination with ground truth when testing the system.

Format: 

[audio file (string)][tab][scene label (string)][tab][original audio file (string)]

Custom setups

If not using the provided cross-validation setup, pay attention to the segments extracted from same original recordings. Original recording identifier can be found from audio segment file names in the evaluation setup\evaluate.txt: 

[original_recording_identifier][start sec][end sec].wav

Make sure that all files having same identifier are placed on the same side of the evaluation.

3. Changelog

1.0 / 2016-11-10

4. License

See file EULA.pdf