Objective voice analysis is vital for diagnosing voice diseases. Traditional vocal assessment parameters, such as jitter, have remained relatively unchanged over the past few decades. Cepstrum technology offers new ways to explore novel vocal assessment parameters. We aimed to explore new parameters applicable to objective voice analysis.A total of 209 patients were analyzed, including 140 patients with hoarseness and 69 healthy controls. Traditional voice characteristics were extracted, and mel-frequency cepstral coefficient (MFCC) analysis was performed. Ten MFCC quantitative parameters were created by statistical analysis. The differences in these MFCC parameters between the two groups were compared. Machine learning algorithms were used to build models, and the importance of each feature was further analyzed on the basis of Shapley Additive exPlanations (SHAPs).There were significant differences between the groups in MFCC-Mean, MFCC-variance (MFCC-Var), MFCC-standard deviation (MFCC-Std), MFCC-75th percentile range, MFCC-maximum value, MFCC-Median, MFCC-skewness, and MFCC-kurtosis values. The values of these parameters in the hoarseness group were significantly greater than those in the normal group. The accuracy, sensitivity, specificity, and the values of the area under the receiver operating characteristic (ROC) curve (AUC) of the hoarseness prediction model established using only the 8 MFCC parameters mentioned above were 95.2%, 94.7%, 96.3%, and 95.5%, respectively, which were only slightly lower than those of the model established using traditional voice assessment indicators, with accuracy, sensitivity, specificity, and AUC values of 96.4%, 96.4%, 96.3%, and 96.4%, respectively. This finding indicates that voice assessment can be performed successfully using only these new MFCC indicators. SHAP analysis revealed that MFCC-Mean, MFCC-Var, and MFCC-Std were the most important predictors of voice disorders.This study is the first to explore novel parameters applicable to objective vocal analysis and shows that parameters such as MFCC-Mean, MFCC-Var, and MFCC-Std are expected to become new voice evaluation indicators.Copyright © 2025. Published by Elsevier Inc.