Flexible wearable vibration optical sensor for voice recognition enhanced property by a microfiber interferometer

Area of Science:

• Biomedical Engineering
• Optical Sensing
• Artificial Intelligence

Background:

• Voice monitoring is crucial for healthcare and authentication.
• Existing sensors may lack flexibility, sensitivity, or accuracy.
• Need for advanced wearable sensors for real-time voice analysis.

Purpose of the Study:

• To develop a flexible wearable optical sensor for voice monitoring.
• To integrate the sensor with an artificial neural network for accurate voice identification.
• To assess the sensor's sensitivity, flexibility, and interference rejection capabilities.

Main Methods:

• Fabrication of a polydimethylsiloxane (PDMS) packaged wearable optical sensor with a microfiber ring.
• Integration with an artificial neural network for voice signal processing.
• Cascading a fiber Bragg grating (FBG) for matrix decoupling of axial strain and temperature interference.
• Training and testing the neural network with sensor-collected speech waveform data.

Main Results:

• The sensor achieved a vibration sensitivity of -0.0316 nm/Hz (100 Hz to 900 Hz).
• Matrix decoupling effectively minimized interference from axial strain (-0.0018 με) and temperature (-1.7286 nm/℃).
• The neural network model successfully recognized different words with 98.33% accuracy.

Conclusions:

• The proposed flexible wearable optical sensor offers high sensitivity and accuracy for voice monitoring.
• The sensor demonstrates potential for applications in voice control, healthcare, and authentication.
• The combination of optical sensing and AI provides a robust solution for biomedical voice analysis.