A Hybrid Fuzzy Soft Prediction Model for Transforming Acoustic Voice Data into Linguistic Knowledge

This paper demonstrates the diagnostic modeling of 20 patients denoted as a universe set U = {x1, x2, …, x20} and its parameter set denoted as E = {VPI, FF, A}, where “VPI” represents the Vocal Perturbation Index, “FF” represents the Fundamental Frequency of the patient, and “A” represents Age. Mathematical prediction models are designed and applied using two well-established mathematical frameworks called Fuzzy Set Theory, which deals with degrees of membership in modeling uncertainty and vagueness in the interval [0,1], and Soft Set Theory, a parametric approach to modeling vagueness. A mapping F: E → F(U) is constructed for the analysis of uncertainty in vocal health assessment, where F(U) denotes the complete lattice of fuzzy subsets of U. In this paper, patients voice recordings are obtained, acoustic features are extracted and embedded in a structured dataset. A triangular membership function is applied in the formulation of the mathematical prediction model to define fuzzy partitions over the parameter space. Here, each crisp data point is converted to corresponding degrees of membership in predefined linguistic classes through the process of fuzzification, and the resulting mathematical inferences provide a rigorous foundation for subsequent approximate reasoning and the decision-support process in vocal health diagnostic assessment.