An AI-Driven Mobile App for IELTS Speaking
Phan, Hong (2025)
2025
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2025112429281
https://urn.fi/URN:NBN:fi:amk-2025112429281
Tiivistelmä
This thesis presents the development of SpeakMate, a mobile application designed to assist IELTS learners (International English Language Testing System) in improving their speaking proficiency through self study, specifically targeting Band 5 to 6 users. The project encompasses the design, implementation, evaluation, and testing of a React Native application integrated with Firebase for user authentication, data storage, and progress tracking.
The system integrates OpenAI Whisper for real-time speech recognition and basic natural language processing (NLP) to transcribe users’ spoken responses (speech to text), generate confidence scores, and provide timestamps for real time feedback and estimated band results. These data points are analyzed to evaluate fluency (e.g., speech rate, pauses) and pronunciation (e.g., clarity) in alignment with IELTS Band 5 6 descriptors.
The thesis includes an overview of challenges in IELTS Speaking preparation, a review of AI driven language learning tools, detailed system design, implementation of key features, an analysis of system limitations, potential future improvements, and cost estimation and monetization methods for larger user tiers.
Results demonstrate that SpeakMate successfully validates the end to end learning workflow for IELTS Speaking Task 1, establishing a stable architectural foundation and enabling the generation of quantitative performance metrics such as Words Per Minute (WPM) and Pause Frequency.
The system integrates OpenAI Whisper for real-time speech recognition and basic natural language processing (NLP) to transcribe users’ spoken responses (speech to text), generate confidence scores, and provide timestamps for real time feedback and estimated band results. These data points are analyzed to evaluate fluency (e.g., speech rate, pauses) and pronunciation (e.g., clarity) in alignment with IELTS Band 5 6 descriptors.
The thesis includes an overview of challenges in IELTS Speaking preparation, a review of AI driven language learning tools, detailed system design, implementation of key features, an analysis of system limitations, potential future improvements, and cost estimation and monetization methods for larger user tiers.
Results demonstrate that SpeakMate successfully validates the end to end learning workflow for IELTS Speaking Task 1, establishing a stable architectural foundation and enabling the generation of quantitative performance metrics such as Words Per Minute (WPM) and Pause Frequency.