Assessment of von Mises-Bernoulli Deep Neural Network in Sound Source Localization

Katsutoshi Itoyama, Yoshiya Morimoto, Shungo Masaki, Ryosuke Kojima, Kenji Nishida, Kazuhiro Nakadai

This paper addresses the properties and effectiveness of the von Mises-Bernoulli deep neural network (vM-B DNN), a neural network capable of learning periodic information, in sound source localization. The phase, which is periodic information, is an important cue in sound source localization, but typical neural network cannot handle periodic input values properly. The vM-B DNN has been theoretically revealed to be able to handle periodic input values and its effectiveness has been shown in a simple case study of sound source localization using artificial sinusoids, but it was not in the case of speech signals. We conducted both numerical simulation and actual environment experiments. We compared a sound source localization method using vM-B DNN with those using ordinary neural networks, and showed that the vM-B DNN outperforms other methods under various conditions.

doi: 10.21437/Interspeech.2021-1050

Cite as: Itoyama, K., Morimoto, Y., Masaki, S., Kojima, R., Nishida, K., Nakadai, K. (2021) Assessment of von Mises-Bernoulli Deep Neural Network in Sound Source Localization. Proc. Interspeech 2021, 2152-2156, doi: 10.21437/Interspeech.2021-1050

@inproceedings{itoyama21_interspeech,
  author={Katsutoshi Itoyama and Yoshiya Morimoto and Shungo Masaki and Ryosuke Kojima and Kenji Nishida and Kazuhiro Nakadai},
  title={{Assessment of von Mises-Bernoulli Deep Neural Network in Sound Source Localization}},
  year=2021,
  booktitle={Proc. Interspeech 2021},
  pages={2152--2156},
  doi={10.21437/Interspeech.2021-1050}
}