Traditional solid-based sensors made from piezoelectric materials or metals have their own boundaries, not limited to problems like poor skin conformity, noise interference, and low sensitivity due to rigidity. As these sensors rely on material deformation or vibration induced by sound pressure, they are less effective in noisy environments and susceptible to unwanted disturbances caused by movements.
To overcome the limitation posed by traditional solid sensors, researchers at Beijing Forestry University have reported a self-filtering liquid acoustic sensor for voice recognition and better human-machine interaction. Researchers claim that the liquid acoustic sensor can improve voice recognition accuracy by 99%, even in noisy surroundings.
According to the study, this remarkable accuracy is achieved by integrating the liquid acoustic sensors with the machine learning algorithms.
“We use the liquid acoustic sensor – together with a machine learning algorithm – to create a wearable voice recognition system that offers a recognition accuracy of 99% in a noisy environment.”, says the study.
Related:
- Microring Resonators could be an efficient solution for advanced computing
- Scaled-up LLMs are more prone to sensible yet wrong answers
- This 5D Memory Crystal can store 360 TB of data for billions of years
These sensors are based on a reconfigurable magnetic liquid called permanent fluidic magnet (PFM) with high remanent magnetization. This means the liquid behaves like a permanent magnet. This magnet contains a three-dimensional oriented and ramified magnetic (ORM) network structure made of neodymium–iron–boron nanoparticles suspended in a carrier fluid.
The 3D ORM network structure enhances the interaction between magnetic particles, boosting the liquid’s magnetic properties.
The study claims the liquid acoustic sensor has a better advantage over solid-based sensors. For instance, The PFM conforms more effectively to the skin, improving contact and sensitivity. Combined with AI, it also recognizes low-frequency sound (less than 30 Hz). With its high sensitivity, the material responds to frequencies from 30 Hz to 10 kHz with high accuracy.
The esteemed liquid acoustic sensor could make its way to wearable devices or fitness trackers to have more control over voice recognition. As PFM can recognize low-frequency sounds, its potential application can be as a hearing aid for users with hearing loss.
Journal Reference:
Zhao, X., Zhou, Y., Li, A., Xu, J., Karjagi, S., Hahm, E., Rulloda, L., Li, J., Hollister, J., Kavehpour, P., & Chen, J. (2024). A self-filtering liquid acoustic sensor for voice recognition. Nature Electronics, 1-9. DOI: 10.1038/s41928-024-01196-y