With Acoustic Signal Processing becoming an area of intensive research, the desire to compare different algorithms calls for comprehensively annotated data collections that are universally accessible and provide all the users with the necessary information for optimizing their algorithms. For many classes of acoustic signal processing tasks this implies the specification of a large number of parameters. Especially for multichannel algorithms, detailed knowledge of spatial parameters (e.g., array topologies, acoustic scene parameters) is usually indispensible for effective signal processing. With the attached generic specification we want to provide a framework for annotating data collections that should be usable for developing and comparing algorithms addressing popular acoustic signal processing tasks such as acoustic echo cancellation, beamforming, and blind source separation.
It is hoped that the readers find it useful and if so, reference it as follows:
W. Kellermann, S. Makino, P. A. Naylor, M. Omologo. (2010). The AcouSP recommendation for annotation of acoustic data collections. (Version 1.0). [On-line]. Available: www.commsp.ee.ic.ac.uk/~acousp.
Moreover, it is hoped that authors of data collections using AcouSP make it publicly available and notify us so that we establish a link.
The authors wish to acknowledge the encouragement by the Technical Committee for Audio and Acoustic Signal Processing (AASP, formerly ‘Audio and Electroacoustics’ (AE)) of the IEEE Signal Processing Society and support by the EC-funded project SCENIC under contract number 226007. They also wish to acknowledge the comments by E. Habets.