Abstract: The present invention discloses a method for producing parametric sound using parametric sound system which is based on ultrasonic electrostatic transducers. It comprises modulation of a carrier ultrasonic signal with a processed audio signal in audio signal processor comprising adaptive frequency filtering based on the audio signal level, dynamic range compression, square root operation, amplification of the modulated ultrasonic signal using a D-class amplifier, driving an electrostatic transducer and generating modulated ultrasonic waves into the air. The electrostatic transducer for the parametric sound system comprises a specific back plate structure that improves electromechanical efficiency of the transducer and also enables realization of a phased array on a single back plate. The disclosed manufacturing method of the electrostatic transducer comprises producing sets of electrodes on the surface of the back plate forming individual cells.
Type:
Grant
Filed:
August 3, 2018
Date of Patent:
August 22, 2023
Assignee:
UAB Neurotechnology
Inventors:
Osvaldas Putkis, Gailius Vanagas, Marius Mikolajunas, Darius Virzonis
Abstract: Disclosed is a system and method for rapid noise-robust friction ridge impression minutiae extraction from digital signal using fully convolutional feed-forward neural network. The proposed neural network based system outperforms classical approaches and other neural network based systems for minutiae extraction in both speed and accuracy. The minutiae extracted using the system can be used at least for tasks such as biometric identity verification, identification or dactyloscopic analysis.