Page 197

www.rsisinternational.org

ILEIID 2025 | International Journal of Research and Innovation in Social Science (IJRISS)

ISSN: 2454-6186 | DOI: 10.47772/IJRISS

Special Issue | Volume IX Issue XXV October 2025

AI-Driven Listening Hub: Enhancing ESL Listening Skills Using

ChatGPT & TTS

Rafidah Amat,

Nur Syazwanie Mansor,

Nor Asni Syahriza Abu Hassan,

Zawani Badri,

Mas Aida

Abd Rahim,

Muhammad Shyazzwan Ibrahim Brian

1,2,3,4,5,6

Academy of Language Studies, Universiti Teknologi MARA, Kedah Branch, Kedah

*Corresponding Author

DOI:

https://dx.doi.org/10.47772/IJRISS.2025.925ILEIID000037

Received: 23 September 2025; Accepted: 30 September 2025; Published: 05 November 2025

ABSTRACT

The invention aims to generate customised listening practices for ESL learners. The artificial intelligence

namely ChatGPT and Text-To-Speech website were selected to create a list of listening practices. With the

emerging trends of the influences of artificial intelligence, it is undeniably in need of intellectual usage. The

innovation lies in its adaptability and accessibility where it enables educators to prompt ChatGPT to produce

an exclusive script-based content and the level of proficiency and later these scripts will be converted into

audio using TTS tools. These listening practices offer a wide range of resources for non-native speakers to

practice their listening skills. Apart from that, the privilege to use these artificial intelligence tools enable the

educators to cater their listening practices to suit students’ level of proficiency. The key features and

functionalities of these products are, it offers educators to create listening practices based on their students;

proficiency, the listening practices can be catered according to the students’ CEFR level as required by the

university or institution. Ever since the demand of self-regulated and scaffolding listening practices is growing,

this tool is beneficial to higher education settings. In fact, it promotes inclusivity and cost-effectiveness where

it empowers educators to create targeted, engaging materials without relying solely on commercial platforms.

Thus, this innovation redefines how educators’ developing and deliver the listening practices, aligning with

modern pedagogical needs in language education.

Keywords: listening practices, listening skills, ChatGPT, Text-To-Speech

INTRODUCTION

Listening comprehension, still, is among the most difficult skills for ESL students, especially in the tertiary

settings such as universities in which the proficiency requirement is in accordance with the internationally

recognized benchmarks e.g. the CEFR. With the swift development of AI, educators now have the chance to

develop responsive and learner-focused listening activities. In particular, we develop an AI-Driven listening

hub, which merges the use of both ChatGPT and Text-to-Speech (TTS) in order to make personalised and

accessible listening materials available in an affordable manner. Using AI, the hub produces text-based scripts

specific to learners’ proficiency levels, which are then transformed into audio by TTS tools in order to foster

inclusive learning and learner agency (Kohnke et al., 2023). Thus, this paper provides a detailed overview of

the system design from generating ChatPGT script to generating the audio from TTS.

Problem Statement

Common listening materials are mostly based on commercial platforms which cannot adapt to learners'

differences. Pre-recorded listening software tends to be inflexible, expensive and not tailored to individual

proficiency levels. ESL Tertiary students find it difficult to access self-regulated and scaffolded practices in

keeping with their tertiary expectations. Teachers, however, are limited in their ability to source diverse,

authentic content that is relevant to a specific student level without the added burdens of the cost of purchasing

and the time it takes to source authentic listening content - such as these guided listening tasks. This calls for

Page 198

www.rsisinternational.org

ILEIID 2025 | International Journal of Research and Innovation in Social Science (IJRISS)

ISSN: 2454-6186 | DOI: 10.47772/IJRISS

Special Issue | Volume IX Issue XXV October 2025

creative solutions that would strike a balance between flexibility, affordability, and pedagogical congruence,

resonating with recent voices advocating for AI-mediated listening practices (Chou&Lee, 2021).

Objectives

1. To design and develop an AI-Driven Listening Hub that integrates ChatGPT and TTS technologies

for ESL listening practices.

2. To generate customised listening scripts based on learners’ CEFR proficiency levels.

3. To convert the generated scripts into audio resources using TTS tools for listening practice.

4. To evaluate the pedagogical effectiveness of the AI-generated practices in promoting listening

comprehension and learner autonomy.

5. To explore the potential of the hub for implementation in higher education institutions.

PRODUCT DESCRIPTION & METHODOLOGY

Figure 1 The AI-Driven Listening Hub

Product Description

Figure 1 presents the AI-Driven Listening Hub screenshots. The hub is an online platform created to provide

personalised listening activities for ESL learners, particularly within the Malaysian context. The hub generates

content through ChatGPT, where educators can specify themes, CEFR levels, and vocabulary focus to produce

relevant scripts. These scripts are then converted into audio using a text-to-speech (TTS) system, resulting in

natural-sounding listening materials that align with learners’ proficiency levels. For instance, the prompts used

to generate text scripts are described based on learners’ proficiency level, a specific context and choosing

accent that best fit learners’ comprehension ability. This ensures more natural delivery as well as assist learners

in comprehending the contents. The completed scripts and audio files are stored within the hub, offering

educators an organised resource for lesson delivery and practice.

METHODOLOGY

Figure 2 The Development Process

Page 199

www.rsisinternational.org

ILEIID 2025 | International Journal of Research and Innovation in Social Science (IJRISS)

ISSN: 2454-6186 | DOI: 10.47772/IJRISS

Special Issue | Volume IX Issue XXV October 2025

Figure 2 describes the development process of AI-Driven Listening Hub. The process started with Design

Prompt to include all the necessary items in generating the required text. Then, the researcher generated the

text using the prompt. The scripts were then be analysed and assessed by two ESL (English as a Second

Language) lecturers to ensure its relevance to the CEFR B1 level. Once it completed, the text was uploaded to

the TTS website to generate the audio. The audio was produced based on Malaysian or Singaporean accent.

Some of these websites do not provide Malaysian accent but only with Singaporean accent. Thus, the

researcher selected either Malaysian or Singaporean accent options. The methodology of this study is designed

to explore ESL learners’ perspectives on the AI-Driven Listening Hub. The development cycle consists of two

interconnected stages: (1) design and development of the hub, and (2) evaluation through learners’ perceptions

of its usability, engagement, and pedagogical value.

Phase 1: Design and Development

Figure 3 The ChatGPT Prompts

Figure 4 The Script

Page 200

www.rsisinternational.org

ILEIID 2025 | International Journal of Research and Innovation in Social Science (IJRISS)

ISSN: 2454-6186 | DOI: 10.47772/IJRISS

Special Issue | Volume IX Issue XXV October 2025

Figure 5 The Script Conversion to TTS website

In this phase, educators prepared listening materials by generating scripts using ChatGPT shown in Figure 3.

Prompts in Figure 4, were designed to specify CEFR levels, targeted vocabulary, and relevant contexts

(academic, social, or professional). These scripts were then converted into audio files in Figure 5, through a

text-to-speech (TTS) system. This TTS was selected based on Malaysian or Singaporean accent as to assist

learners’ familiarisation as well as better understanding of the contents delivered. The final materials (both

scripts and audio) were integrated into the hub and organised for structured use by learners.

Phase 2: Learners’ Perspectives

The second phase focused on collecting ESL learners’ perceptions of the hub. A group of students from higher

education institutions were introduced to the platform and assigned listening tasks based on the generated

materials. Data collection employed surveys and open-ended questions to capture their views on usability,

accessibility, and engagement. Emphasis was given to how learners experienced scaffolding, inclusivity, and

autonomy within the platform. Qualitative feedback provided insights into the perceived effectiveness of the

hub in supporting listening development and meeting their academic needs.

Potential Findings and Commercialisation

The AI-Driven Listening Hub will help students improve their listening skills as it will offer individualized and

level-appropriate student tasks. It will also contribute to self-regulated learning and scaffolding, both of which

will lead to learner autonomy and to academic achievement (Oxford, 2017). It is expected that learners will

benefit from synthetic listening materials, notwithstanding that some challenges in terms of robotic tone and

lack of prosody and technological issues continue to exist (Choi, 2022; Kang et al., 2009). In order to mitigate

these issues, the audios can be altered according to its speed, tone as well as the pitch to lessen the robotic style

of the audio which hinders learners’ understanding.

Table 1 First Question of Open-ended

Themes

Examples of Responses

Technical Issues (internet, lagging,

connection problems, downloading)

S17: “Must have strong Network connection.”

S20: “No Internet cause difficulty to me as a student.”

Pronunciation & Accent Problems

S1: “AI cannot understand emotion.”

S10: “The intonations … are quite monotonous.”

Page 201

www.rsisinternational.org

ILEIID 2025 | International Journal of Research and Innovation in Social Science (IJRISS)

ISSN: 2454-6186 | DOI: 10.47772/IJRISS

Special Issue | Volume IX Issue XXV October 2025

S29: “Unnatural accent.”

Lack of Emotional / Natural

Expression

S22: “…difficult to understand because of the tone sounds unnatural and it lacks

emotions.”

S23: “AI didn’t… pick up subtle cues like hesitation, sarcasm, or passive-

aggressiveness.”

Speed and Clarity Issues

S13: “its maybe speak fast.”

S22: “…sometimes the AI talks too fast or uses words I don’t know.”

S43: “The speed level is very high may be difficult to some listeners.”

Content Accuracy / Quality

S4: “AI-generated listening tools have errors sometimes.”

S41: “Accuracy Issues: AI may misinterpret accents, slang, or technical terms.”

S54: “misinformation.”

Accessibility / Cost Barriers

S17: “Must have strong Network connection.”

S30: “need to purchase the pro.”

No Challenges

S5: “Nothing.”

S40: “I think it’s nothing.”

S48: “so far I didn’t face any of it.”

The findings show that the AI-Driven Listening Hub offers innovative opportunities for ESL listening practice,

but learners also reported two main types of challenges. The first are technical issues, such as unstable internet

connections, lagging systems, and subscription costs. The second are linguistic issues, including robotic or

unclear pronunciation, unnatural accents, lack of emotion, and fast speech. These limitations are consistent

with earlier studies that highlight weaknesses in speech synthesis, particularly in prosody and cultural nuance

(Choi, 2022; Kang et al., 2009). At the same time, some learners indicated that they faced no difficulties,

suggesting that AI-generated materials can work well for students with strong internet access or those more

adaptable to synthetic voices. This shows that while AI-based tools are useful, learners value the authenticity

of human voices, especially the natural rhythm, intonation, and emotional expression that support deeper

listening comprehension (Susilo, 2023; Nainggolan & Hanifah, 2024). Overall, the results suggest that AI

listening tools should be used as a complement rather than a replacement for traditional resources. A blended

approach would balance the accessibility and flexibility of AI with the authenticity and cultural richness of

human-based listening practices (Back & Kabulis, 2025; Jung, 2025).

NOVELTY AND RECOMMENDATIONS

The novelty of this innovation lies in its dual adaptability and inclusivity. Unlike static commercial resources,

the hub empowers educators to create dynamic, script-based listening materials tailored to learners’

proficiency levels. It integrates two key AI functionalities which are ChatGPT for generating authentic,

context-rich scripts and TTS for producing natural-sounding audio resources This AI-Driven Listening Hub

provides learners with scaffolding, learner autonomy as well as cost-effective access to listening practices. The

hub is also accessible through mobile phones in which it provides convenience to the learners. Future research

should examine how blending AI-generated content with human-delivered listening practices can maximise

authenticity and pedagogical impact (Wang & Vasquez, 2023). It is recommended that the project undergo

pilot testing within a higher education setting to validate its effectiveness. Furthermore, continuous refinement

should be pursued through feedback loops involving educators and learners. Future developments may explore

multilingual expansions, integration with mobile learning applications, and incorporation of analytics features

to track learner progress. Last but not least, this listening hub can be expanded to include diverse ESL

populations as well as the tracking to learners’ progress.

ACKNOWLEDGEMENTS

The authors would like to extend their deepest appreciation to the Akademi Pengajian Bahasa (APB),

Universiti Teknologi MARA (UiTM) Kedah Branch, for their continuous support and encouragement

throughout the development of this project. Sincere thanks are also extended to the students involved in the

Page 202

www.rsisinternational.org

ILEIID 2025 | International Journal of Research and Innovation in Social Science (IJRISS)

ISSN: 2454-6186 | DOI: 10.47772/IJRISS

Special Issue | Volume IX Issue XXV October 2025

study, who feedback and engagement significantly informed the process of innovation and a special thanks to

research team members who dedicated their best ideas, labour, and abilities to this project.

REFERENCES

1. Back, M., & Kabulis, K. (2025). Modality or Authenticity?: The Role of Authentic versus AI-Generated

Texts in Language Learner Engagement. In Rethinking Language Education in the Age of Generative AI

(pp. 68-89). Routledge.

2. Birtchnell, T. (2018). Listening without ears: Artificial intelligence in audio mastering. Big Data &

Society, 5(2), 2053951718808553.

3. Jung, H. (2025). AI-Assisted Student-Generated Listening Materials in High School EFL: A Sociomaterial

Perspective. Multimedia-Assisted Language Learning, 28(2), 55-76.

4. Nainggolan, E. E., & Hanifah, H. (2024). Listening Comprehension Problems Encountered by EFL

Students at Coastal Area. Teaching and Learning Journal of Mandalika (Teacher) e-ISSN 2721-9666, 5(2),

208-217.

5. Raza, M. A., Khan, H., & Bukhari, S. (2024). Transforming EFL Listening Skills: The Power of AI

Integration in Classrooms. Social Science Review Archives, 2(2), 2284-2295.

6. Susilo, J. (2023). An Analysis Of Students Difficulties In Using Authentic Recording In Listening Skill Of

The Tenth Grade On SMK Citra Angkasa Bandar Lampung (Doctoral dissertation, IAIN Metro).

7. Trang, M. (2020). Understanding listening comprehension processing and challenges encountered:

Research perspectives. International Journal of English Language and Literature Studies, 9(20), 63-75