Reconstruction of multiple-user voice commands using a hybrid system based on thin-film electronics and CMOS

This paper presents a system consisting of an array of thin-film microphone channels on glass, which can be formed on large substrates. Each microphone channel consists of a polyvinylidene difluoride (PVDF) piezoelectric transducer as well as amplifier and scan circuits based on amorphous-silicon (a-Si) thin-film transistors (TFTs). The scan circuits multiplex signals from multiple channels to a CMOS IC for readout. By spatially distributing the channels on a large substrate, audio signals from multiple simultaneous speakers in a space can be both acquired in closer proximity and separated, enabling a multi-user human-computer interface based on voice commands. To overcome low TFT performance in the scan circuits (which limits channel sampling to below the Nyquist rate), a signal reconstruction algorithm is proposed. An 8-channel system demonstrates acquisition and reconstruction of 2 simultaneous audio signals at 2m distance from the array.