EXPLORING ENGLISH STUDENTS’ PERCEPTIONS OF GPT-BASED VOICE FEATURES IN ENHANCING SPEAKING PROFICIENCY

Abstract

Speaking is an essential component of English language learning, yet many EFL learners still experience difficulties in developing fluency and confidence in oral communication. Although the use of Artificial Intelligence (AI), particularly ChatGPT, has been widely discussed in language education, research focusing on its voice feature as a tool for speaking practice remains limited. This study explores English students’ perceptions of using the GPT-based voice feature to support the development of speaking skills. A quantitative survey approach was adopted, involving 48 English Education students at Universitas Islam Indragiri selected through purposive sampling. Data were gathered using a five-point Likert-scale questionnaire addressing three aspects of perception: absorption, understanding, and evaluation. The data were analyzed descriptively using percentages and mean scores. The results indicate that students generally held positive perceptions of the voice feature across all indicators. Absorption (173%), Understanding (171%), and Evaluation (168%), with a total mean score of 171, indicating a strong overall perception. They found the feature easy to use, engaging, and helpful in improving pronunciation, fluency, and speaking confidence. These findings suggest that the GPT-based voice feature can function as an effective supplementary medium for speaking practice outside the classroom. Overall, this study provides practical insights into the integration of AI voice technology in EFL speaking instruction and contributes to the growing discussion on innovative approaches to language learning.