Showing papers on "Transducer published in 2021"

Developing Real-Time Streaming Transformer Transducer for Speech Recognition on Large-Scale Dataset

Abstract: Recently, Transformer based end-to-end models have achieved great success in many areas including speech recognition. However, compared to LSTM models, the heavy computational cost of the Transformer during inference is a key issue to prevent their applications. In this work, we explored the potential of Transformer Transducer (T-T) models for the fist pass decoding with low latency and fast speed on a large-scale dataset. We combine the idea of Transformer- XL and chunk-wise streaming processing to design a streamable Transformer Transducer model. We demonstrate that T-T outperforms the hybrid model, RNN Transducer (RNN-T), and streamable Transformer attention-based encoder-decoder model in the streaming scenario. Furthermore, the runtime cost and latency can be optimized with a relatively small look-ahead.

Design and Fabrication of a Novel Dual-Frequency Confocal Ultrasound Transducer for Microvessels Super-Harmonic Imaging

Abstract: Recently, super-harmonic ultrasound imaging technology has caused much attention due to its capability of distinguishing microvessels from the tissues surrounding them. However, the fabrication of a dual-frequency confocal transducer is still a challenge. In this work, 270- $\mu \text{m}$ PMN-PT single crystal 1–3 composite and 28- $\mu \text{m}$ PVDF thick film, acting as transmission layer and receiving layer, respectively, are integrated in a novel co-focusing structure. To realize delicate wave propagation control, microwave transmission line theory is introduced to design such structure. Two acoustic filter layers, 13- $\mu \text{m}$ copper layer and 39- $\mu \text{m}$ Epoxy 301 layer, are indispensable and should be added between two piezoelectric layers. Therefore, an acoustic issue can be overcome via an electrical method and the successful achievement of a dual-frequency (5 MHz/30 MHz) ultrasound transducer with a confocal distance of 8 mm can be realized. The super-harmonic ultrasound imaging experiment is conducted using this kind of device. The 3-D image of 110- $\mu \text{m}$ -diameter phantom tube injected with microbubbles can be obtained. These promising results demonstrate that this novel dual-frequency (5 MHz/30 MHz) confocal ultrasound transducer is potentially usable for microvascular medical imaging application in the future.

A Wearable Ultrasound System for Sensing Muscular Morphological Deformations

Abstract: Noninvasive monitoring of muscle contraction, which provides information about muscle morphological deformations, has a great potential in medical applications, such as stroke rehabilitation and prosthesis control This paper presents a wearable multichannel A-mode ultrasound system for the multiperspective muscle contraction detection against its existing bulky ultrasound sensing counterpart The system consists of a waveform generator and a waveform amplifier for ultrasound excitation, as well as a signal processing module for echo receiving, amplifying, and transmitting In addition, a miniaturized transducer was optimized for muscle contraction monitoring using 1–3 piezoelectric composite The system’s superiorities on excitation pulse, detection depth, and axial resolution were validated by the evaluation experiments And in vivo muscle deformation detection and virtual prosthesis control experiments (target achievement control test) demonstrated its ability in rehabilitation applications, with a task completion rate of 100% and path efficiency of 9330% These results confirmed the reliability of the proposed ultrasound system and paved the way for its applications in rehabilitation treatment and prosthesis research

Multiple Solutions of the Tristable Energy Harvester

Abstract: This paper presents the results of numerical simulations of a non-linear, tristable system for harvesting energy from vibrating mechanical devices. Detailed model tests were carried out in relation to the system consisting of a beam and three permanent magnets. Based on the derived mathematical model and assuming a range of control parameter variability, a three-dimensional image of the distribution of the largest Lyapunov exponent was plotted. On its basis, the regions of chaotic and predictable movement of the considered system exist have been established. With reference to selected plane of the largest Lyapunov exponent cross-sections, possible co-existing solutions were identified. To identify multiple solutions, a diagram of solutions (DS) diagram was used to illustrate the number of existing solutions and their periodicity. The proposed calculation tool is based on the so-called fixed points of Poincare cross-section. In relation to selected values of the control parameter ω, coexisting periodic solutions were identified for which phase trajectories and basins of attraction were presented. Based on the model tests carried out, it was found that in order to efficiently harvest energy, appropriate transducer adjustment is required. Calibration of the transducer is necessary to obtain the greatest amplitude of vibration of the beam, which corresponds to the phase trajectory limited by external energy potential barriers. As expected, the average voltage induced on the electrodes of the piezoelectric transducer and the average electrical power recorded on the resistive element are directly proportional to the amplitude and average kinetic energy of the beam.

Deep Learning for Ultrasound Beamforming in Flexible Array Transducer

Abstract: Ultrasound imaging has been developed for image-guided radiotherapy for tumor tracking, and the flexible array transducer is a promising tool for this task. It can reduce the user dependence and anatomical changes caused by the traditional ultrasound transducer. However, due to its flexible geometry, the conventional delay-and-sum (DAS) beamformer may apply incorrect time delay to the radio-frequency (RF) data and produce B-mode images with considerable defocusing and distortion. To address this problem, we propose a novel end-to-end deep learning approach that may alternate the conventional DAS beamformer when the transducer geometry is unknown. Different deep neural networks (DNNs) were designed to learn the proper time delays for each channel, and they were expected to reconstruct the undistorted high-quality B-mode images directly from RF channel data. We compared the DNN results to the standard DAS beamformed results using simulation and flexible array transducer scan data. With the proposed DNN approach, the averaged full-width-at-half-maximum (FWHM) of point scatters is 1.80 mm and 1.31 mm lower in simulation and scan results, respectively; the contrast-to-noise ratio (CNR) of the anechoic cyst in simulation and phantom scan is improved by 0.79 dB and 1.69 dB, respectively; and the aspect ratios of all the cysts are closer to 1. The evaluation results show that the proposed approach can effectively reduce the distortion and improve the lateral resolution and contrast of the reconstructed B-mode images.

A Rotary Ultrasonic Motor Operating in Torsional/Bending Modes With High Torque Density and High Power Density

Abstract: In this article, to achieve high torque density and high power density of rotary ultrasonic motors (USMs), we develop a rod-shaped transducer operating in torsional/bending (T/B) modes and excite two elliptical motions on its bilateral ends to drive the rotor orthogonally pressed onto the transducer. Here, the torsional vibration, the driving structure, and the rod shape may enable the T/B motor to provide high driving force, large driving distance, and low weight, which facilitate the realization of high torque density and high power density. To assess the validity, first, we construct a transducer 43 and 110 mm in diameter and length, respectively. Then, we investigate the load characteristics of the T/B motor: When it works at the frequency of 21.64 kHz, the maximal torque and the maximal output power reach 10.1 N·m and 38.1 W, respectively. In the meantime, it yields the torque density and the power density of, respectively, 20.0 N·m/kg and 75.7 W/kg, which are, as predicted, relatively high compared to the values of most conventional rotary motors. This article not only validates the effectiveness of our proposal but also provides a new approach to design powerful rotary USMs.

Development of Frequency-Mixed Point-Focusing Shear Horizontal Guided-Wave EMAT for Defect Inspection Using Deep Neural Network

Abstract: We propose a new frequency-mixed point-focusing shear horizontal (SH) guided-wave electromagnetic acoustic transducer (EMAT) in this work to obtain the defect positions and plate thickness simultaneously and accurately. Compared with other guided-wave detection methods, it is not required to measure the plate thickness in advance because we can easily obtain it during the test. We use the variational mode decomposition method to decompose the received frequency-mixed defect signal into subsignals with different center frequencies and to remove the noise. Furthermore, we use the continuous wavelet transform to analyze these subsignals using the time–frequency method and to obtain the time-of-flight information of the guided wave under different frequencies and modes. Therefore, we can obtain accurate defect positions and plate thicknesses via the new transducer and signal processing methods while improving the signal intensities. In the identification of defect types, we first constructed a database set containing three types of defects of different sizes using data enhancement methods. Then, the dense network, convolutional neural network, recurrent neural network, and newly proposed deep GFresNet are studied to analyze the defect classification performance of each structure. The results show that the proposed GFresNet has very good defect identification accuracy, which is about 95% along any depth of the defects, and that it can automatically extract high-level information without sophisticated feature engineering.

Ultrahigh-Gain Organic Electrochemical Transistor Chemosensors Based on Self-Curled Nanomembranes

Abstract: Organic electrochemical transistors (OECTs) are technologically relevant devices presenting high susceptibility to physical stimulus, chemical functionalization, and shape changes-jointly to versatility and low production costs. The OECT capability of liquid-gating addresses both electrochemical sensing and signal amplification within a single integrated device unit. However, given the organic semiconductor time-consuming doping process and their usual low field-effect mobility, OECTs are frequently considered low-end category devices. Toward high-performance OECTs, microtubular electrochemical devices based on strain-engineering are presented here by taking advantage of the exclusive shape features of self-curled nanomembranes. Such novel OECTs outperform the state-of-the-art organic liquid-gated transistors, reaching lower operating voltage, improved ion doping, and a signal amplification with a >104 intrinsic gain. The multipurpose OECT concept is validated with different electrolytes and distinct nanometer-thick molecular films, namely, phthalocyanine and thiophene derivatives. The OECTs are also applied as transducers to detect a biomarker related to neurological diseases, the neurotransmitter dopamine. The self-curled OECTs update the premises of electrochemical energy conversion in liquid-gated transistors, yielding a substantial performance improvement and new chemical sensing capabilities within picoliter sampling volumes.

Tiny Transducer: A Highly-Efficient Speech Recognition Model on Edge Devices

Abstract: This paper proposes an extremely lightweight phone-based transducer model with a tiny decoding graph on edge devices. First, a phone synchronous decoding (PSD) algorithm based on blank label skipping is first used to speed up the transducer decoding process. Then, to decrease the deletion errors introduced by the high blank score, a blank label deweighting approach is proposed. To reduce parameters and computation, deep feedforward sequential memory network (DFSMN) layers are used in the transducer encoder, and a CNN-based stateless predictor is adopted. SVD technology compresses the model further. WFST-based decoding graph takes the context-independent (CI) phone posteriors as input and allows us to flexibly bias user-specific information. Finally, with only 0.9M parameters after SVD, our system could give a relative 9.1% - 20.5% improvement compared with a bigger conventional hybrid system on edge devices.

Living Rat SSVEP Mapping with Acoustoelectric Brain Imaging.

Abstract: Acoustoelectric Brain Imaging (ABI) is a potential method for mapping brain electrical activity with high spatial resolution (millimeter). To resolve the key issue for eventual realization of ABI, testing the hypothesis that recorded acoustoelectric (AE) signal can be used to decode intrinsic brain electrical activity, the experiment of living rat SSVEP measurement with ABI is implemented. Method A 1-MHz ultrasound transducer is focused on the visual cortex of anesthetized rat. With visual stimulus, the electroencephalogram and AE signal are simultaneously recorded with Ag electrode. Besides, with FUS transducer scanning at the visual cortex, corresponding AE signals at different spatial positions are decoded and imaged. Results Consistent with that of direct measurement of SSVEP, the decoded AE signal presents a clear event-related spectral perturbation (ERSP). And, the decoded AE signal is of high amplitude response at the base and harmonics of the visual stimulus frequency. Whats more, for timing signal, a significant positive amplitude correlation is observed between decoded AE signal and simultaneously measured SSVEP. In addition, the mean SNRs of SSVEP and decoded AE signal are both significantly higher than that of background EEG. Finally, with one fixed recording electrode, the active area with an inner diameter of 1mm is located within the 4mm4mm measurement region. Conclusion These experimental results demonstrate that the millimeter-level spatial resolution SSVEP measurement of living rat is achieved through ABI for the first time. Significance This study confirms that ABI should shed light on spatiotemporal resolution neuroimaging.

An Electrical and Ultrasonic Doppler System for Industrial Multiphase Flow Measurement

Abstract: Accurate measurement of flow parameters of multiphase flow has great significance for safe, economic, and efficient operation of the industrial production process. To simultaneously measure the water holdup and velocity, a combined electrical and ultrasonic Doppler measurement system based on a compact peripheral component interconnect (CPCI) bus is designed. The combined sensor consists of a six-ring conductance sensor, two pairs of capacitance plate sensors, and an ultrasonic Doppler sensor. The measurement system is cored with a field-programmable gate array (FPGA) and consists of a series of functional modules. Through flexible programming and interboard communication, the system can meet different measurement requirements. To meet industrial data transmission standards and increase system extensibility, the demodulated data are transmitted to a host computer through the CPCI bus for further data analysis. Dynamic experiments demonstrated that the multimodal system has stable and preferable performance in water holdup and velocity measurement under all flow patterns. The conductance and capacitance sensor could realize the water holdup measurement with the root-mean-square error (RMSE) of 3.02% and the overall superficial velocity estimation through cross correlation with the RMSE of 0.09 m/s. Besides, the pulse ultrasonic Doppler transducer could measure the overall superficial velocity through the area integration of the velocity profile with the RMSE of 0.08 m/s.

Airborne power ultrasound for drying process intensification at low temperatures: Use of a stepped-grooved plate transducer

Abstract: The recent development of a novel family of power ultrasonic transducers with extensive radiating surfaces represents a step forward in the implementation of new airborne power ultrasonic (APU) tec...

Fabrication and high acoustic performance of high frequency needle ultrasound transducer with PMN-PT/Epoxy 1-3 piezoelectric composite prepared by dice and fill method

Abstract: In this paper, we reported the fabrication and high acoustic performance of high frequency needle ultrasound transducer based on 0.70Pb(Mg1/3Nb2/3)O3-0.30PbTiO3 single crystal/Epoxy 1–3 piezoelectric composite, which was prepared by using the dice and fill (DAF) method. The as-prepared 0.70PMN-0.30 PT/Epoxy 1–3 composite has high thickness extension mode electromechanical coupling coefficient with the value of 67.3 %. The transducer is designed by PMN-PT/Epoxy 1–3 piezoelectric composite with a dimension of 550 μm × 450 μm × 35 μm. The fabricated transducer has not only a high center frequency larger than 42 MHz, but also high bandwidth of -6 dB of 90.6 %, the insertion loss with the value of 16 dB and excellent transmitting sensitivity of 314 mV/V. The results are in good agreement with the values simulated by PiezoCAD software. In addition, the acoustic performance of the transducer was also obtained by measuring the acoustic filed distribution in purified water. The -3 dB acoustic beam dimension width, the value of focal length and the acoustic power were 0.38 mm × 0.42 mm, 2.3 mm and 4.4 × 10−3mW, respectively. The performance PMN-PT/Epoxy 1–3 piezoelectric composite ultrasound transducer fabricated by dice and fill method (DAF) achieved the effect of ultrasonic transducer fabricated by deep reactive ion etching (DRIE). The excellent acoustic performance demonstrated that the DAF prepared PMN-PT/Epoxy 1–3 piezoelectric composite have great industry application potential in high frequency ultrasonic electronics, such as microelectronics NDT and biomedical imaging.

Development of Broadband High-Frequency Piezoelectric Micromachined Ultrasonic Transducer Array

Abstract: Piezoelectric micromachined ultrasonic transducers (PMUT) are promising elements to fabricate a two-dimensional (2D) array with a pitch small enough (approximately half wavelength) to form and receive arbitrary acoustic beams for medical imaging. However, PMUT arrays have so far failed to combine the wide, high-frequency bandwidth needed to achieve a high axial resolution. In this paper, a polydimethylsiloxane (PDMS) backing structure is introduced into the PMUTs to improve the device bandwidth while keeping a sub-wavelength (λ) pitch. We implement this backing on a 16 × 8 array with 75 µm pitch (3λ/4) with a 15 MHz working frequency. Adding the backing nearly doubles the bandwidth to 92% (−6 dB) and has little influence on the impulse response sensitivity. By widening the transducer bandwidth, this backing may enable using PMUT ultrasonic arrays for high-resolution 3D imaging.

Unidirectional Shear Horizontal Wave Generation by Periodic Permanent Magnets Electromagnetic Acoustic Transducer With Dual Linear-Coil Array

Abstract: Shear horizontal (SH) waves are commonly generated by periodic permanent magnet (PPM) electromagnetic acoustic transducers (EMATs) in metallic media. Conventional PPM EMATs generate ultrasonic waves, which simultaneously propagate both forward and backward. This can be an undesirable characteristic, since the backward wave can be eventually reflected, reaching the receiver transducer where it can mix with the signal of interest. This limitation can be overcome using two side-shifted PPM arrays and racetracks coils to generate SH waves in a single direction. That design relies on the EMAT wavefront diffraction to produce constructive and destructive interference, but produces unwanted backward traveling sidelobes. Here, we present a different design, which uses a conventional PPM array and a dual linear-coil array. The concept was numerically simulated, the main design parameters were assessed and the unidirectional EMAT was experimentally evaluated on an aluminum plate, generating the SH0-guided wave mode nominally in a single direction. The amplitude ratio of the generated waves at the enhanced to the weakened side is above 20 dB. Since the wavefronts from the two sources are perfectly aligned, no obvious backward sidelobes are present in the acoustic field, which can significantly reduce the probability of false alarm of an EMAT detection system.

Embedded piezoelectric transducers based early-age hydration monitoring of cement concrete added with accelerator/retarder admixtures:

Abstract: Accelerator/retarder admixtures are often added into concrete to improve its early-age strength, which needs to be effectively monitored during its hardening process. The electromechanical impedanc...

Piezoelectric Scanning Micromirror With Built-In Sensors Based on Thin Film Aluminum Nitride

Abstract: A micromirror with piezoelectric thin film aluminum nitride (AIN) as transducer material for actuation and detection is presented. Four sensor elements are integrated for closed-loop monitoring of the resonant driven microscanner with an entire footprint of 6 mm2. The sensor and actuator elements are monolithically fabricated in 150 mm SOI technology. The micromirror is characterized in terms of its scanning and sensor characteristics. A large optical scan angle of 137.9° (34.5° mechanical tilt angle corresponding) at 3385 Hz and 20 V actuation voltage is achieved. The median resonant voltage sensitivity is 4.7 °/V. For the sensor characterization, a cross-talk compensation is introduced, which significantly improves the signal quality and thus enables an accurate charge determination. The sensor signal increases linearly to the deflection. A dynamic sensor angle sensitivity of 48.4 fC/° is reached. Possible applications of the micromirror are consumer electronics, such as LIDAR or medical applications, such as fluorescence microscopy.

Lead-Free KNN-Based Textured Ceramics for High-Frequency Ultrasonic Transducer Application

Abstract: Environment-friendly lead-free piezoelectric materials with excellent piezoelectric properties are needed for high-frequency ultrasonic transducer applications. Recently, lead-free 0.915(K0.45Na0.5Li0.05)NbO3-0.075BaZrO 3-0.01(Bi0.5Na0.5)TiO3 (KNLN-BZ-BNT) textured piezo- electric ceramics have high piezoelectric response, superior thermal stability, and excellent fatigue resistance, which are promising for devices applications. In this work, the KNLN-BZ-BNT textured ceramics were prepared by the tape-casting method. Microstructural morphology, phase transition, and electrical properties of KNLN-BZ-BNT textured ceramics were investigated. High-frequency needle-type ultrasonic transducers were designed and fabricated with these textured ceramics. The tightly focused transducers have a center frequency higher than 80 MHz and a −6-dB fractional bandwidth of 52%. Such transducers were built for an ${f}$ -number close to 1, and the desired focal depth was achieved by press-focusing technology associated with a set of customer design fixture. Its lateral resolution was better than $90~\mu \text{m}$ by scanning a 15- $\mu \text{m}$ tungsten wire target. These promising results demonstrate that the lead-free KNLN-BZ-BNT textured ceramic is a good candidate for high-frequency ultrasonic transducer applications.