Engineering study demonstrates noise-robust throat motion and vocal monitoring using flexible piezoelectric arrays, indicating potential for wearable dysphagia and speech diagnosis.
Key Points
Develop a flexible, noise-resistant piezoelectric sensor array for continuous monitoring of vocal cord vibrations and laryngeal movements to diagnose speech disorders and dysphagia.
Fabricated screen-printed lead zirconate titanate (PZT) nanocomposites using a heterogeneous matrix with an optimal 4:6 bisphenol A to bisphenol F epoxy ratio on a 20 µm polyimide base.
Attached the 60 µm-thick multi-channel patch to the neck to measure laryngeal movement during swallowing and acoustic performance under extreme background noise.
Integrated convolutional neural networks to classify four distinct voice states from the acquired vibration signals.
The sensor achieved a piezoelectric coefficient (d₃₃) of 105 pC/N, sensitivity of 5.80 mV/kPa, crack resistance at a 0.5 mm bending radius, and <3.5% output deviation under a 5 mm radius.
Tracked laryngeal elevation accurately (24 ± 1 mm for saliva; 33 ± 1 mm for water) and captured vocal vibrations with an SNR of 32.06 dB under hair dryer noise, outperforming standard microphones by 29.97 dB.
Machine learning classification using convolutional neural networks achieved 95.93% accuracy across four voice states during continuous all-day activity.