Contact Piezoelectric Transducers for Acoustic Voice Analysis: High Reliability for Frequency and Intensity, but...

Clinical voice assessment is often limited by the environmental sensitivity of air-conduction microphones. While contact transducers offer noise immunity, their validity across diverse clinical parameters remains debated. This study evaluates the interchangeability of a dual-element piezoelectric transducer against a professional headset microphone, investigating the impact of anatomical filtering on signal integrity.

Forty-nine adults (VHI-10 < 5.5) underwent simultaneous recordings. To account for biomechanical filtering, comprehensive anthropometric profiling (neck circumference and cervical skinfold thickness) was performed. Statistical validity was assessed via linear mixed-effects models (LMM) and intraclass correlation coefficients (ICC).

The transducer achieved superior noise rejection (SNR: 28.2 dB vs 24.8 dB). Mean fundamental frequency exhibited excellent interchangeability (ICC = 0.839, P = 0.909), and intensity metrics showed moderate reliability. In contrast, spectral (Slope, Tilt), noise (HNR), and amplitude perturbation (Shimmer) measures revealed significant systematic bias (P < 0.05). Furthermore, all perturbation (Jitter, Shimmer) and noise-related metrics exhibited poor reliability (ICC < 0.30), precluding their individual clinical use. Crucially, anthropometric variables did not emerge as significant covariates, suggesting that anatomical variability, while present, does not systematically improve sensor performance models. These in vivo findings challenge previous synthetic models, highlighting the damping characteristics of neck tissue.

Low-cost piezoelectric transducers are robust tools for monitoring pitch and intensity in ecologically valid, noisy environments. However, they remain unsuitable for clinical perturbation analysis due to biomechanical low-pass filtering. This study provides a necessary methodological framework and calibration equations for integrating contact sensors into ambulatory voice parameter tracking and telepractice.