Voice-driven Self-powered Ethanol Sensor for Speech Recognition and Drunk Driving Detection

Abstract

The integration of multisensory capabilities into flexible and wearable platforms is necessary but remains challenge for intelligent human–machine interfaces and personal health monitoring. Herein, we reported a multifunctional active smart textile (MAST) based on electrospun PVDF/TiO₂ nanofibers for simultaneous acoustic recognition and ethanol detection. The MAST leverages the piezoelectric activity of a PVDF core scaffold to achieve highly sensitive acoustic detection, covering a wide range of sound pressure level from 60 to 97 dB with an exceptional frequency resolution of up to 1 Hz. Meanwhile, a TiO₂ shell layer was decorated on the surface of PVDF framework to provide highly selective ethanol discrimination. The devices not only accurately distinguish the vocal features among versatile living creatures with a recognition rate over 95%, but also efficiently identify various biomotions like joint bending and pressing. More importantly, the interfacial coupling of acoustoelectric core and chemoresistive shell enables autonomous encoding of exhaled ethanol into output vocal print signal with excellent linearity and reversibility. By integrating Mel-Frequency Cepstral Coefficients (MFCCs) with Dynamic Time Warping (DTW) algorithm, we developed a MAST enabled intelligent vehicle system (MIVS) capable of precisely and actively detecting speaker’s identity and degree of intoxication at the same time. This work offers a new possibility of voice-driven self-powered chemical quantification, paving the way for development of next-generation wearable intelligent textiles.