Frequency coverage improved flexible PMN-PNN-PZT acoustic sensor via gradual PI film for noisy environments speech recognition

Abstract

As it progresses to a hyperconnected society in the era of the Fourth Industrial Revolution, acoustic sensors are in the spotlight as the most intuitive and free-hands interface through speech recognition in human-machine interaction. Speech recognition in a noisy noise environment is important for such human-machine interaction, but it is difficult to increase accuracy with a single acoustic sensor due to its single channel and low sensitivity. Therefore, in this paper, a flexible piezoelectric acoustic sensor that mimics the basilar membrane of the cochlea, a human auditory organ, is produced, and sensitivity is improved using resonance and voice information is increased in a single sensor using multiple channels. For accurate speech recognition, sensitivity improvement in the human speech range of 100 Hz to 8000 Hz is implemented as a polyimide substrate that mimics the change in the thickness of the basilar membrane. In order to improve accuracy through higher sensitivity, PNN-doped PMN-PZT thin film is manufactured by sol-gel coating, which is transferred to a gradual thickness polyimide substrate to realize a flexible piezoelectric acoustic sensor. The manufactured gradual flexible piezoelectric acoustic sensor compared voltage and sensitivity characteristics with the previous flat flexible piezoelectric acoustic sensor, and compared speech recognition rate with commercial microphones in noisy noise environments. This paper is meaningful in that speech recognition in a noise environment is possible in a single sensor by improving the performance of an acoustic sensor.