Rancang Bangun Aplikasi Latihan Pengucapan Bahasa Inggris Berbasis Speech Recognition dengan Metode Sorensen-Dice Coefficient

English pronunciation is a critical yet challenging aspect of language learning, particularly in Indonesia where access to native speakers and real-time feedback is limited. This research develops an Android-based English pronunciation training application utilizing speech recognition technology integrated with the Sorensen-Dice Coefficient method to provide objective and measurable feedback. The system captures user speech, converts it into text via speech recognition, and evaluates pronunciation accuracy by calculating textual similarity between the spoken transcription and reference text, resulting in a percentage-based score. The application was developed using the Rapid Application Development (RAD) methodology with Kotlin, Jetpack Compose, PHP, and MySQL. Testing involved 15 respondents performing 270 pronunciation trials across six beginner-level topics. Results show an overall accuracy of 71.9%, classified as “Fair” based on Gorunescu’s (2011) grading standard. The application successfully delivers instant, structured feedback, making it a practical self-learning tool for beginner English learners