Robust speech enhancement against speaker's movements in noisy reverberant environments for hearing aid

2015 Fiscal Year Final Research Report

• Project/Area Number: 25350670
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Rehabilitation science/Welfare engineering
• Research Institution: Oita University
• Principal Investigator: Furuya Ken'ichi 大分大学, 工学部, 教授 (10643611)
• Co-Investigator(Renkei-kenkyūsha): NISHIJIMA Keisuke 大分大学, 工学部, 助教 (30237698)
• Project Period (FY): 2013-04-01 – 2016-03-31
• Keywords: マイクロホンアレイ / 画像処理 / 非負値行列因子分解 / 音響モデル

Outline of Final Research Achievements

We propose two speech enhancement methods for hearing aid. The first is the method combining image information captured by camera and audio information captured by microphone array. Experiment results show that the performance with the proposed method using microphone array and camera is superior to the system only with microphone array. The second method uses the Multichannel Nonnegative Matrix Factorization (MNMF) for speech enhancement. The problem of MNMF is initial-value dependency for the performance. In experimental results, the initial-value settings based on the phase differences between the microphones was more effective than the random settings

Free Research Field

音響信号処理