Comparative Acoustic Analysis in Dental Plosives Using Linear Predictive Coding for Articulation Disorder Assessment

Abstract

Dental plosives /d/ and /t/ in Kannada (Mysore dialect) share nearly identical place and manner of articulation, making them acoustically similar and particularly difficult for articulation-disordered individuals to learn. This study presents a comparative acoustic analysis establishing a baseline profile from a normal native speaker, with each phoneme iterated five times and analysed using Praat for spectrogram generation and Origin 2026b for statistical visualization. Formant frequencies (F1–F4) were compared across three estimation approaches—direct measurement, Linear Predictive Coding (LPC), and Short-Time Fourier Transform (STFT) envelope peaks—alongside intensity, pitch, jitter, and shimmer. Results show jitter and shimmer remain nearly identical between the two phonemes, while intensity and pitch maxima differ significantly, offering a reliable acoustic marker for distinguishing them. LPC yielded smoother, more accurate resonance tracking than the comparatively noisy STFT envelope, especially at lower frequencies. These baseline findings provide a quantitative acoustic reference supporting diagnostic and therapeutic strategies in speech-language pathology.