• Cindy Owusu Tabiri

  Author

Artificial Intelligence (AI) is a technology that has proven to be valuable to improve English speaking through automated feedback. It is an effective method to encourage students to learn English by themselves and enhance their communication abilities automatically, evaluating and providing support for individual learning. But a lot of current AI-powered speaking systems are mainly geared towards transcribing the speech or awarding points for pronunciation. However, several existing systems that use artificial intelligence technology for speaking purposes do not provide sufficient feedback in an educational context, focusing mostly on speech recognition or pronunciation evaluation. This paper presents an Artificial Intelligence-Based Automated Feedback Framework for English Speaking Practice that merges the potential of automatic speech recognition (ASR) technology with that of a large language model (LLM). The framework provides feedback on pronunciation, fluency, grammar, vocabulary usage, and speaking performance. The proposed framework was evaluated on the dataset of English speech recognition from Kaggle. To process the speech signals, noise reduction, silence removing, and normalization were performed, and then speech signals were transcribed by the Whisper ASR model. Learner-centered feedback and suggestions for improvement were given through the process of speaking feature extraction and analysis. For evaluating the performance, the data set was divided into 70% as training, 15% as validation, and 15% as testing data. The experimental results showed that 94.2% of pronunciation accuracy score, 91.6% of fluency score, 90.4% of grammar improvement score, 89.8% of vocabulary improvement score and 93.1% of feedback acceptance rate, which proved that the proposed framework is effective in autonomous English-speaking practice. In total, the framework had an average educational performance score of 91.8% for the five assessment items, indicating its performance as a tool for giving students all-round formative feedback. The proposed approach is an intelligent and scalable solution for the use of AI for English language learning in academic settings.

• PDF