Ultra-wide frequency response and high-resolution triboelectric acoustic sensors for constructing multifunctional speech aided system

Abstract

As an important medium for human-computer interaction, acoustic sensors can directly convey human intentions and provide information. However, it is challenging for existing acoustic sensors to simultaneously achieve a broad spectrum and high-resolution response to speech, which is crucial for accurately achieving human beings’ smart speech recognition for human-computer interaction in artificial intelligence (AI) era. In this work, we fabricate a high-performance triboelectric acoustic sensor (HPTS) equipped with an ultra-wide band response range, capable of collecting human speech across the entire range of audible frequencies from 20 Hz to 20,000 Hz, exhibiting ultra-high frequency resolution down to 0.1 Hz. Subsequently, a multifunctional speech aided system is constructed using HPTS that is capable of effectively achieving the recognition of subtle emotional nuances, semantics and voiceprint. Enhanced by the emotional semantic recognition model (ESRM), this speech aided system has achieved a much higher accuracy in recognizing seven different human emotions, reaching an impressive accuracy of 95.43 %. Additionally, it is proficient in interpreting various semantics and voiceprint from human users. This speech aided system with ultra-wideband response and ultra-high frequency resolution provides a highly promising solution for future human-computer interaction and has a wide range of potential applications in AI.