Sharpening Directivity of Beamforming Based on Neural Networks Using Virtual Observations
Sharpening Directivity of Beamforming Based on Neural Networks Using Virtual Observations
复制标题
使用虚拟观测的基于神经网络的波束形成锐化方向性
DOI：
10.1109/gcce53005.2021.9621789
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
Mitsuhiro Mizumachi,?Ryoichi Miyazaki
中科院分区：
文献类型：
--
作者：
Yuki Kanekiyo;Mitsuhiro Mizumachi,?Ryoichi Miyazaki
基于神经网络的波束形成是一种非线性波束形成方法。这种方法可以在麦克风较多的情况下提高性能，但由于硬件限制，可能无法轻松增加麦克风的数量。因此，我们将重点放在虚拟观察上，这是仅使用实际观察对在麦克风实际放置位置处观察到的信号的估计。本文提出将虚拟观测应用于基于神经网络的波束形成。通过评估实验，我们证实了所提出的方法可以获得比仅使用实际麦克风的情况更清晰的方向性。
Beamforming based on neural networks (NN) has been proposed as nonlinear beamforming. This method improves the performance with more microphones, but it may not be possible to easily increase the number of microphones due to hardware restrictions. Therefore, we focus on virtual observation, which is an estimation of the signal observed at a location where the microphone is not actually placed, using only actual observations. In this paper, we propose to apply virtual observation to NN-based beamforming. From the evaluation experiment, we confirmed that the proposed method can obtain sharper directivity than the case of using only actual microphones.