An aerosol jet-printed wearable graphene/cellulose nanocrystal acoustic sensor for speech recognition

Abstract

Wearable acoustic sensors offer a promising solution for effective communication for individuals with speech impairments by calibrating throat vibrations and converting them to synthesized speech. We developed a new type of piezoresistive acoustic sensor that is additively manufactured with polyurethane (PU) film-encapsulated graphene/cellulose nanocrystals (CNCs) via aerosol jet printing. Featuring a high degree of biocompatibility and flexibility, the sensor is capable of accurately measuring varying sound pressure levels (SPLs). Results have demonstrated that the acoustic sensitivity of the sensor can be regulated by changing the graphene concentration, and the sensor with the 20 wt. % of graphene concentration manifests the high sensitivity of 9.7 × 10–6 dB–1, a broad working range from 30 to 90 dB, and a minimum resolution for SPL change of 10 dB, along with a linear correlation between SPL and the resistance change measured by the sensor. When adhered to the throat of a study participant as a wearable device, the sensor faithfully captures subtle vocal characteristics including timbre and rhythm. Integrated with a support vector machine (SVM)-based machine learning algorithm, the device achieves a high accuracy of 95.9% when used to recognize digits (0–9), assisting people with speech difficulties to communicate in a digital manner.