Generalized Singular Value Decomposition for Personalized Audio Using Loudspeaker Array

Gauthier PA, Pasco Y, Berry A. Generalized Singular Value Decomposition for Personalized Audio Using Loudspeaker Array. In: AES Conference: 2016 AES International Conference on Sound Field Control. Audio Engineering Society; 2016. Paper 1-2. Available from: https://aes.org/publications/elibrary-page/?id=18302

@inproceedings{Gauthier2016_18302,
  author = {Gauthier, Philippe-Aubert and Pasco, Yann and Berry, Alain},
  title = {{Generalized Singular Value Decomposition for Personalized Audio Using Loudspeaker Array}},
  booktitle = {AES Conference: 2016 AES International Conference on Sound Field Control},
  note = {Paper 1-2},
  year = {2016},
  month = jul,
  publisher = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=18302}
}

TY  - CPAPER
TI  - Generalized Singular Value Decomposition for Personalized Audio Using Loudspeaker Array
AU  - Gauthier, Philippe-Aubert
AU  - Pasco, Yann
AU  - Berry, Alain
T2  - AES Conference: 2016 AES International Conference on Sound Field Control
M1  - Paper 1-2
PY  - 2016
DA  - 2016/07/06
UR  - https://aes.org/publications/elibrary-page/?id=18302
PB  - Audio Engineering Society
LA  - en
AB  - Personalized audio is the creation of independent sound zones. The zones are distinguished as the bright zone and the dark zone. The desired audio signal should be audible in the bright zone and reduced in the dark zone. Known methods are the pressure matching method, acoustic contrast maximization, beamforming, and high-pass filtering of cylindrical harmonic expansions. Several challenges are related to personalized audio, including the sound pressure level difference between bright and dark zones. This paper presents a theoretical investigation of a new potential method to achieve personalized audio: Generalized singular value decomposition of multichannel transfer matrices for the automatic creation of source distributions that independently operate on each zone. Results of simulations provide convincing results.
ER  -