Related Experiment Video
Updated: May 29, 2025
07:23
Fabrication of Nanoheight Channels Incorporating Surface Acoustic Wave Actuation via Lithium Niobate for Acoustic Nanofluidics
Published on: February 5, 2020
5.7K
Triboelectric Nanogenerator-Based Flexible Acoustic Sensor for Speech Recognition
Yang Dai1, Yunlong Li1, Shixian Xuan1
1School of Mechanical and Electrical Engineering, Xi'an Polytechnic University, Xi'an 710048, China.
ACS Applied Materials & Interfaces
|February 6, 2025
Summary
Researchers developed a novel flexible acoustic sensor using a sound-driven triboelectric nanogenerator. This innovative sensor achieves high sensitivity and accuracy for human-machine interaction and speech recognition applications.
Area of Science:
- Materials Science
- Electrical Engineering
- Acoustics
Background:
- Flexible acoustic sensors are crucial for advancing human-machine interaction (HMI).
- Existing sensors face challenges in balancing cost, stability, fidelity, and sensitivity.
- Triboelectric nanogenerators (TENGs) offer potential for novel acoustic sensing solutions.
Purpose of the Study:
- To propose and develop a low-cost, high-performance flexible acoustic sensor based on a sound-driven TENG.
- To integrate this sensor with a deep learning model for effective speech recognition.
- To demonstrate the sensor's capabilities in HMI applications.
Main Methods:
- Fabrication of a PVDF/GO composite nanofiber film as the dielectric layer via electrospinning.
- Utilization of copper-nickel alloy conductive fabric as electrodes.
- Design of an imitation embroidery shed structure for sensor assembly.
- Development of a multilayer attention convolutional network (MLACN) for speech recognition.
Main Results:
- The fabricated sensor achieved a high sensitivity of 4.76 V·Pa-1.
- A wide frequency response range of 20-2000 Hz was demonstrated.
- The integrated speech recognition system attained 99.5% accuracy for common word pronunciations.
Conclusions:
- The proposed sound-driven TENG-based flexible acoustic sensor offers a promising solution for HMI.
- The combination of advanced materials and structural design enhances sensor performance.
- Integration with deep learning (MLACN) enables highly accurate speech recognition, paving the way for future HMI advancements.
