CHiME-3 Background Noise
The CHiME-3 background noise dataset contains continuous recordings from four everyday environments: buses, cafés, pedestrian areas and street junctions. The recordings come from the Third CHiME Speech Separation and Recognition Challenge and can be used to create noisy speech mixtures for research and benchmarking.
The release contains 102 WAV files from 17 recording sessions. Each session was recorded with a six-microphone tablet array, with one 16 kHz WAV file per channel. This release covers the background recordings only. The speech recordings and transcriptions are part of the separate full CHiME-3 distribution.
The background recordings are available under a CC BY 4.0 licence. Download the CHiME-3 background noise dataset from Hugging Face.
Citation
If you use these recordings, please cite both the Zenodo dataset record and the CHiME-3 challenge paper.
- Barker, J., Marxer, R., Vincent, E., and Watanabe, S. (2015). The Third ‘CHiME’ Speech Separation and Recognition Challenge: Dataset, Task and Baselines. IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU). doi:10.1109/ASRU.2015.7404837
@inproceedings{barker2015chime3,
author = {Jon Barker and Ricard Marxer and Emmanuel Vincent and Shinji Watanabe},
title = {The Third {CHiME} Speech Separation and Recognition Challenge: Dataset, Task and Baselines},
booktitle = {2015 IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU)},
year = {2015},
doi = {10.1109/ASRU.2015.7404837}
}