by N. Bertin1, E. Camberlein1, R. Lebarbenchon1, E. Vincent2, S. Sivasankaran2, I. Illina2 and F. Bimbot1
1IRISA - CNRS UMR 6074, Rennes, France
2Inria, Villers-lès-Nancy, F-54600, France
This corpus includes reverberated, noisy speech signals spoken by 12 native French talkers in 4 houses (3 rooms per house) and recorded by an 8-microphone device at various angles and distances and in various noise conditions.
This corpus stands apart from other corpora in the field by the number of rooms and homes considered by the diversity of acoustic conditions recorded and by the facts that it is publicly available at no cost.
VoiceHome-2 corpus and its documentation can be downloaded here.
We provide here baseline software and tools to perform source localization, speech enhancement (via source separation) and automatic speech recognition on the corpus. These baseline scripts allow to reproduce the results presented in the publication cited below.
The baseline scripts use these two related toolboxes, also downloadable from their respective homepages.
All participants have given an informed and signed consent about public diffusion of recorded sentences.