Development of a superdirective microphone-array system based on the soft computing technology

2014 Fiscal Year Final Research Report

• Project/Area Number: 24560519
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Measurement engineering
• Research Institution: University of Yamanashi
• Principal Investigator: OZAWA Kenji 山梨大学, 総合研究部, 教授 (30204192)
• Co-Investigator(Kenkyū-buntansha): KINOSHITA Yuichiro 山梨大学, 大学院総合研究部, 准教授 (70452133)
• Research Collaborator: ISEKI Akihiro ; AKISHIKA Yusuke ; MORISE Masanori
• Project Period (FY): 2012-04-01 – 2015-03-31
• Keywords: マイクロホンアレイ / ニューラルネットワーク / 遺伝的アルゴリズム / ソフトコンピューティング / 指向性 / 時空間音圧分布 / GPU / 最適化

Outline of Final Research Achievements

This study deals with the technology of spatial recording of acoustical signals and aims to improve the spatial resolution by introducing the soft computing technology.
The optimization of a neural-network-based microphone-array system was achieved by determining the system parameters and the neural-network-learning parameters using a genetic algorithm. Then the system was expanded to learn super-directivities for multiple frequencies. As a result, the super-directivities for sinusoidal signals with various frequencies were yielded. Finally, we developed a wideband system in which an input complex tone is decomposed into sinusoidal components using the fast Fourier transform, and processed in parallel with independent array-processing units.

Free Research Field

音響情報工学