To overcome signal distortion in real-life noisy environments, Prof. Lee's group integrated the multi-channel f-PAS with advanced deep learning-based speech processing algorithms. The group demonstrated a noise-robust flexible piezoelectric acoustic sensor (NPAS) that leverages a novel Deep U-net based Speech Enhancement (DEEP-SEA) model alongside convolutional neural network (CNN) classifiers. This approach enables exceptional speech enhancement and speaker recognition, dramatically reducing error rates even under severe thermal, environmental, or crowd-noise conditions. 
For commercial mobile applications and AIoT devices, the group successfully miniaturized the acoustic sensor into a piezoelectric mobile acoustic sensor (PMAS). By adjusting the internal residual stress and enhancing lateral dipoles of the ultrathin PZT membrane, the PMAS maintains high sensitivity within a scaled-down footprint. This highly sensitive PMAS was fully integrated with a machine learning processor and a customized smartphone application to demonstrate real-time, highly accurate mobile biometric authentication utilizing only a small amount of training data.
[Related References]
“Theoretical Basis of Biomimetic Flexible Piezoelectric Acoustic Sensors for Future Customized Auditory Systems” Adv. Funct. Mater., 34, 2309316, 2024
“Deep learning-based noise robust flexible piezoelectric acoustic sensors for speech processing” Nano Energy, 101, 107610, 2022
“Biomimetic and flexible piezoelectric mobile acoustic sensors with multiresonant ultrathin structures for machine learning biometrics” Sci. Adv., 7, eabe5683, 2021
“Flexible Piezoelectric Acoustic Sensors and Machine Learning for Speech Processing” Adv. Mater., 32, 1904020, 2020
“Basilar membrane-inspired self-powered acoustic sensor enabled by highly sensitive multi tunable frequency band” Nano Energy, 53, 198, 2018
