Conductive fabric diaphragm for noise-suppressive headsetmicrophone

Author

Hiraki, Hirotaka and Kanazawa, Shusuke and Miura, Takahiro and Yoshida, Manabu and Mochimaru, Masaaki and Rekimoto, Jun

Abstract

Voice interaction systems require high-quality speech input; however, ambient noise significantly degrades the performance of conventional microphones. While multimodal sensor-based speech enhancement techniques have been proposed, they typically require GPU devices for processing. We proposed a conductive fabric diaphragm headset microphone, which integrates a planar fabric microphone into a headset configuration. The proposed microphone exhibits low sensitivity and curved directivity characteristics, enabling efficient capture of the user’s voice while simultaneously suppressing background and competing speech noises. Comparative analyses against two commercially available noise-cancelling headsets demonstrated that our approach achieved superior Scale-Invariant Signal-to-Noise Ratio (Si-SNR) performance. Our approach constitutes a real-time, hardware-based noise-suppressive microphone solution with potential applications in speech communication, automatic speech recognition, and voice conversion technologies operating in noise-rich environments.