Probability-Based Best Sample Selection for Acoustic Analysis of Normal and Disordered Voices

Area of Science:

• Speech and Hearing Sciences
• Acoustic Analysis
• Voice Disorders

Background:

• Quantitative vocal fold function assessment relies on accurate acoustic analysis.
• Operator selection of stable voice segments is critical for analysis accuracy.
• Existing methods for stable segment selection may lack reliability.

Purpose of the Study:

• To introduce a novel modal periodogram method for improved voice segment selection.
• To enhance the accuracy and reliability of acoustic analysis in voice assessment.
• To provide a more robust tool for clinicians and researchers analyzing voice function.

Main Methods:

• Four segment selection methods were compared: modal periodogram, moving window, midvowel, and whole vowel.
• Acoustic parameters including jitter, correlation dimension (D2), and spectrum convergence ratio (SCR) were calculated.
• The Wilcoxon Signed-Rank test was used to compare the modal periodogram method against standard techniques.

Main Results:

• The modal periodogram method demonstrated significantly lower D2 values (P < 0.01).
• This method also yielded significantly higher SCR values for both normal and disordered voices (P < 0.01).
• These results indicate superior performance in selecting stable audio segments.

Conclusions:

• The proposed modal periodogram method offers enhanced accuracy and reliability for voice segment selection.
• This technique is more adept than current standard methods for acoustic analysis.
• The findings support the clinical utility of the modal periodogram method in voice assessment.