Audio Based Screening Tool to Identify Speech Sound Disorder – Kannada

Abstract

India’s linguistic diversity highlights the need for language-specific tools for the accurate identification of communication disorders. Speech Sound Disorders (SSD) are common in early childhood and, if not detected early, can lead to long-term academic and social difficulties. However, standardized screening systems for regional languages such as Kannada are limited. This study proposes a web-based Kannada speech screening system for early identification of SSD among children aged 2–7 years in Karnataka. The system is based on the SODA (Substitution, Omission, Distortion, Addition) framework to systematically analyze articulation errors. It incorporates browser-based audio recording, speech-to-text processing, and phonetic transcription using the International Phonetic Alphabet (IPA), followed by rule-based phoneme comparison to detect deviations in speech production. The system generates structured and interpret-able reports highlighting error patterns to support clinical assessment and early intervention. Experimental findings indicate that substitution and omission are the most frequent error types, aligning with typical speech development patterns. The proposed approach improves screening efficiency, objectivity, and accessibility, providing a scalable and practical solution for early detection of SSD in Kannada-speaking children.