A self-filtering liquid acoustic sensor for voice recognition

Area of Science:

• Materials Science
• Acoustics
• Biomedical Engineering

Background:

• Wearable acoustic sensors are crucial for voice recognition.
• Solid-state sensors face limitations due to ambient noise, motion artifacts, and poor skin conformability.

Purpose of the Study:

• To introduce a novel liquid acoustic sensor for enhanced voice recognition.
• To overcome the limitations of traditional solid-state wearable acoustic sensors.

Main Methods:

• Fabrication of a liquid acoustic sensor using a 3D oriented magnetic network of neodymium-iron-boron nanoparticles in a carrier fluid.
• Integration of the liquid acoustic sensor with a machine learning algorithm for voice recognition.
• Characterization of sensor performance including pressure discrimination, signal-to-noise ratio, and motion artifact filtering.

Main Results:

• The liquid acoustic sensor demonstrated high sensitivity, discriminating pressures as low as 0.9 Pa.
• Achieved a high signal-to-noise ratio of 69.1 dB.
• The sensor effectively filtered low-frequency biomechanical motion artifacts (below 30 Hz).

Conclusions:

• A wearable voice recognition system utilizing the liquid acoustic sensor achieved 99% accuracy in noisy conditions.
• The liquid sensor's unique properties offer superior performance over conventional solid-state devices.
• This technology presents a promising advancement for wearable human-machine interfaces.