Performance of Nonlinear Energy Difference Ratio and Voice Type Component Profile Across Phoneme, Relative Sound...

Nonlinear energy difference ratio (NEDR) and voice type component profile (VTCP) are reliable acoustic metrics for voice classification; however, their performance under varying vocal intensities, speech sounds, and recording environments has not been thoroughly established. This study examined the effectiveness of these measures under real-world variability with different phonemes, sound intensities, and environments to help determine their clinical diagnostic potential.

Thirty-one normophonic adults produced sustained vowels (/a/, /i/, /u/) and fricatives (/s/, /z/) at five nominal sound pressure levels from very soft (L1), soft (L2), baseline (L3), loud (L4), and very loud (L5). The vowel /a/ was also recorded in sound-attenuated and noise-masked environments to elucidate the environmental effects on reliability. Trimmed 0.75-second samples were analyzed in MATLAB to compute NEDR and VTCP values. Repeated-measures analysis of variances tested the effects of intensity, sound type, and environment for NEDR and VTCP.

NEDR remained stable across vowels and voiced fricatives, with deviations limited to very soft phonation (L1) and the voiceless fricative /s/. VTCP was more sensitive to changes, particularly at low intensities and in noisy environments, with voice types 3 and 4 frequently misclassifying healthy phonations as disordered. Across environments, NEDR showed greater robustness than VTCP.

NEDR demonstrates relative stability across vocal intensity, phoneme type, and recording environment, supporting its potential clinical utility, including in telepractice and nonsoundproof settings. Its susceptibility to very low intensities and voiceless fricatives highlights the need for cautious interpretation and further validation in disordered populations.