A Comparative Study of Acoustic Feature Extraction Using CSL and Web Prototype of LIS-N Application

The study compared the LIS-N web prototype application, which uses the Praat-Parselmouth library for voice analysis, with the clinical standard Computerized Speech Lab (CSL). Twenty adult volunteers without voice problems completed three voice tasks over three consecutive days. Tasks included reading the Rainbow Passage, sustaining the vowel /i/, and maximum phonation time /a/. Fundamental frequency and pitch mean demonstrated excellent agreement between systems (r > 0.98) and were robust to hardware and software differences. Energy-based features showed moderate agreement, with differences primarily driven by microphone variations. Perturbation measures including jitter, shimmer, and CPP showed consistently poor cross-platform agreement. The study confirms the viability of open-source acoustic tools as accessible alternatives to proprietary clinical systems and supports their use in telehealth and remote voice monitoring.