Association Between Perturbation Analysis of Sustained Vowel Phonation and Cepstral Analysis of Connected Speech

Background and Objectives Traditional acoustic analysis based on sustained vowels does not fully represent voice in daily communication. This study aimed to examine the relationship between perturbation-based acoustic parameters from sustained vowel phonation and cepstral-based parameters from connected-speech tasks in both normal and disordered voices.Materials and Method Voice data were collected from 18 speakers with normal voices and 29 speakers with voice disorders using the Multi-Dimensional Voice Program and the Analysis of Dysphonia in Speech and Voice of the Computerized Speech Lab. Correlations among jitter, shimmer, cepstral peak prominence (CPP), and ratio of low to high-frequency spectral energies (L/H ratio) were examined using Spearman’s rank correlation analysis.Results In the voice disorder group, CPP showed strong negative correlations with jitter (rho=-0.685, p<0.001) and shimmer (rho=-0.651, p<0.001) in the all-voiced sentence, and moderate negative correlations with jitter (rho=-0.596, p=0.001) and shimmer (rho=-0.543, p=0.002) in the sentence including all Korean vowels. CPP and L/H ratio demonstrated strong positive correlations across the two connected-speech tasks (CPP: rho=0.804, p<0.001; L/H ratio: rho=0.777, p<0.001). In the normal group, CPP from the vowel-inclusive sentence showed a moderate negative correlation with jitter (rho=-0.566, p=0.014), while the L/H ratio from both sentences showed moderate negative correlations with shimmer (all-voiced: rho=-0.565, p=0.015; vowel-inclusive: rho=-0.567, p=0.014). Both CPP (rho=0.834, p<0.001) and L/H ratio (rho=0.686, p=0.002) exhibited positive correlations between the two tasks.Conclusion Cepstral analysis using connected-speech tasks complements perturbation-based analysis. In particular, the sentence including Korean all vowels, containing consonants, may serve as a practical connected-speech task reflecting daily voice use in both normal and disordered speakers.