Sound ?eld reproduction (SFR) applied to microphone array recordings can be de?ned as an inverse problem. A typical issue is the resulting spatial source signal distribution. A least-square approach with solution 2-norm regularization will activate all sources. To circumvent such issues, sparsity-based regularization offers promising advantages. However, this can result in too sparse solutions. A promising avenue is based on structured-sparsity for prede?ned groups of reproduction sources. To this end, the group lasso was recently investigated without overlapping group. Its inability to deal with overlapping groups is a limitation. This paper presents the latent group lasso (LGL) where group-level sparsity is induced for overlapping groups. Theoretical results of LGL are successfully compared to other SFR results for sparsity-inducing norms.
Gauthier, Philippe-Aubert; Grandjean, Pierre; Berry Alain
Affiliations: University of Sherbrooke, Sherbrooke, Quebec, Canada; Centre for Interdisciplinary Research in Music, Media, and Technology, McGill University, Montreal, Quebec, Canada(See document for exact affiliation information.)
AES Conference: 2018 AES International Conference on Spatial Reproduction - Aesthetics and Science (July 2018)
Paper Number: P9-1
Publication Date: July 30, 2018
Session Subject: Sound field reproduction; Structured sparsity; Sparsity; Loudspeaker array; Latent Group; Lasso; Elastic-net
No AES members have commented on this paper yet.
If you are not yet an AES member and have something important to say about this paper then we urge you to join the AES today and make your voice heard. You can join online today by clicking here.