A Study on Practical Talker Tracking System

2017 Fiscal Year Final Research Report

• Project/Area Number: 15K06084
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Communication/Network engineering
• Research Institution: Tokyo Denki University
• Principal Investigator: Suyama Kenji 東京電機大学, 工学部, 教授 (50303011)
• Project Period (FY): 2015-04-01 – 2018-03-31
• Keywords: 音源追尾 / マイクロホンアレー / 確率的最適化

Outline of Final Research Achievements

In this research, a method for multiple sound source tracking by using two microphones system was studied. In the two microphone system, an estimation accuracy of the DOAs (Direction-Of-Arrival) tends to degrade due to a low spatial resolution. Therefore, MUSIC (Multiple Signal Classification), which is known as a high resolution method for the spectrum estimation, is applied. Although the MUSIC spectrum function indicates sharp peaks to sound source directions, some false peaks are appeared in the function in the high frequency band when a width between two microphones is extended for improving the estimation accuracy. Thus, the evaluation function becomes to a multi-modal function. In this research, PSO (Particle Swarm Optimization) was applied for the peak search. In addition, the same source tracking problem was also resolved by setting a penalty function.

Free Research Field

信号処理