A Wearable AI‐Driven Mask with Humidity‐Sensing Respiratory Microphone for Non‐Vocal Communication
Jianfei Wang, Hongyu Zhang, Xiaomin Wu, Mingyan Gao, Wen He, Zhibo Zhang, Kremena Makasheva, Wen J. Li, Zuobin Wang
Abstract
Hoarseness and dysphonia caused by vocal cord conditions or laryngeal surgeries significantly hinder communication and quality of life. This study presents a plug-and-play humidity-sensing respiratory microphone (HSRM) with generalized features for individual users. Leveraging gold nanoparticle-based humidity sensors integrated into commercially available wearable face masks, the system enables patients to produce verbal communication without relying on vocal cord activity. By integrating nanoparticle-enhanced humidity sensors with advanced convolutional neural networks, the HSRM system accurately decodes respiratory patterns into intelligible speech, achieving a recognition accuracy of 85.61%. Leveraging nanoparticle-polymer interfaces that effectively convert atmospheric humidity fluctuations into precise electrical signals, the system pioneers a contactless and non-invasive paradigm in assistive speech technology. This innovation addresses limitations of existing devices, such as reliance on residual vocal fold vibrations or skin-contact sensors, offering a practical generalized solution for individuals with aphonia. With its potential to facilitate naturalistic communication and transform healthcare applications, the HSRM system sets a new benchmark in wearable assistive technologies for voice rehabilitation and human-machine interaction.