Development of impulsive noise suppression with real-time processing

2019 Fiscal Year Final Research Report

• Project/Area Number: 17K00272
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Human interface and interaction
• Research Institution: Tottori University
• Principal Investigator: SASAOKA Naoto 鳥取大学, 工学部, 准教授 (80432607)
• Co-Investigator(Kenkyū-buntansha): 伊藤 良生 鳥取大学, 工学研究科, 教授 (70263481)
• Project Period (FY): 2017-04-01 – 2020-03-31
• Keywords: 信号処理 / 騒音抑圧 / 音声強調 / 高次統計量 / ヒアラブルデバイス

Outline of Final Research Achievements

For a hearable device and hearing-aid, noise suppression is a key technology. Besides, these devices require real-time processing with low delay. The impulsive noise converts an initial peak sound to a ringing sound due to the natural mode radiation. The initial peak sound has a wideband component, and the ringing sound is composed of a narrowband component. Unfortunately, a conventional method suppresses only the wideband component. Therefore, we propose to suppress not only the wideband but also the narrowband component. A linear predictor using the least mean fourth algorithm can suppress the narrowband component in noisy speech because the fourth-order moment of speech is about zero in the short-time analysis. Besides, the proposed noise detector, which uses fourth-order cumulant, can detect a wideband noise frame, and then the wideband noise can be suppressed. From computer simulations, we can see the proposed method improves the noise reduction ability.

Free Research Field

ディジタル信号処理

Academic Significance and Societal Importance of the Research Achievements

残響を伴う突発性雑音のモデル化並びに統計的性質を明らかにすることにより、雑音除去システム発展の一助となる。また、ヒアラブルデバイスや補聴器に必須となる低遅延リアルタイム処理可能な雑音除去システムを開発することで、ヒアラブルデバイスの応用範囲を広げ、ひいては難聴の方が生活しやすい環境をつくる一助となる。