Spatial properties of the DEMAND noise recordings

Joachim Thiemann 1 Emmanuel Vincent 2 Steven Van de Par 1
2 PAROLE - Analysis, perception and recognition of speech
Inria Nancy - Grand Est, LORIA - NLPKD - Department of Natural Language Processing & Knowledge Discovery
Abstract : "DEMAND" (Diverse Environments Multichannel Acoustic Noise Database) is a set of recordings of environmental noises in both indoor and outdoor settings. The recordings were performed with a 16-channel planar array of microphones. The purpose of the recording is to provide researchers with a large set of freely available noise recordings (licensed under a Creative Commons licence) for use in developing algorithms such as beamforming, noise reduction, and source separation, although anyone may use the data for any purpose they see fit. A more detailed description of the DEMAND recordings can be found in [1]. In this article, we examine some of the spatial properties of the DEMAND recordings, in particular the cross-channel correlations. Notably, the quality of the reverberation characteristics is compared to the theoretical ideal. This property is used as a post-recording calibration of the microphone positions, compared to the design speci cations of the array.
Type de document :
Communication dans un congrès
40th Annual German Congress on Acoustics (DAGA 2014), Mar 2014, Oldenburg, Germany. 2014
Liste complète des métadonnées

Littérature citée [3 références]  Voir  Masquer  Télécharger
Contributeur : Emmanuel Vincent <>
Soumis le : mercredi 30 avril 2014 - 17:10:15
Dernière modification le : mardi 18 décembre 2018 - 16:38:02
Document(s) archivé(s) le : mercredi 30 juillet 2014 - 14:05:47


Fichiers produits par l'(les) auteur(s)


  • HAL Id : hal-00985979, version 1



Joachim Thiemann, Emmanuel Vincent, Steven Van de Par. Spatial properties of the DEMAND noise recordings. 40th Annual German Congress on Acoustics (DAGA 2014), Mar 2014, Oldenburg, Germany. 2014. 〈hal-00985979〉



Consultations de la notice


Téléchargements de fichiers