Higher Education Category Submission

811 – ChatnLearn: Enhancing Personalised Learning through Speech-to-Speech AI Conversation for Hong Kong Secondary Students

What if learning wasn't just about listening, but about talking it through? ChatnLearn brings this idea to life. Our AI-powered platform uses interactive conversational avatars to help students truly engage with their learning. Inspired by the concept of oracy - the ability to articulate ideas through spoken language - ChatnLearn enhances learning with two key features. 'Talk to Learn' encourages learners to think out loud as they consolidate academic concepts across subjects, while 'Learn to Talk' offers a safe space for students to practice speaking in a second language.

Technically, ChatnLearn leverages a conversational video interface powered by speech recognition and large language models (LLMs). This enables dynamic interactions with avatars who ask questions, challenge thinking, and adapt to students' responses in real time.

The platform addresses major challenges in traditional education, including large class sizes, limited speaking opportunities, and high cost for external support such as private tutoring. By creating a low-pressure environment for speaking and learning at an affordable price, ChatnLearn helps students from all backgrounds to build communication skills and confidence.

The goal of ChatnLearn is to empower students with the articulation skills and confidence to express their thoughts clearly in a foreign language. Ideally, it can be used for lesson preparation, providing students with a relaxed, chit-chatty environment that ultimately enhances their performance not just in exams, but also in real-world situations like presentations, interviews, and future careers.

Hong Kong’s education system is highly exam-oriented. While students often perform well in written tests, they struggle to express ideas clearly when speaking, especially in English. Research shows that Hong Kong students’ English proficiency lags behind peers in Singapore, the Philippines, and Malaysia. It lowers our students’ competitiveness, especially when it comes to interviews, presentations and networking.

Large class size is one of the reasons that limits students' opportunities to practice speaking. With many students in one classroom, teachers cannot give everyone enough time to articulate their thoughts, leading to passive learning habits.

At the same time, education inequality is a growing issue. Students from lower income families often lack access to extra learning support at home, such as exposure to foreign languages or hiring a private tutor. There is a strong need for affordable, personalised solutions that can help all students to perform better regardless of their background.

The idea for ChatnLearn comes from observing these gaps. Our hypothesis is that while Hong Kong students are tech-savvy in using online resources, there is a lack of affordable solutions that focus on oral practice and critical thinking.

ChatnLearn proposes a simple model: students complete one oral practice session per day or per week using our conversational AI avatars, receive instant feedback, and visualise their improvement over time. Many schools in Hong Kong already subscribe to e-learning platforms, so the technology adoption barrier is low. Unlike many existing tools that are chatbot-like, ChatnLearn focuses on real speaking practice without putting extra burden on teachers. In fact, ChatnLearn helps teachers to enhance lesson preparation with student performance analytics, allowing them to better adjust teaching pace and content.

We believe that by providing a second chance for students to practice speaking with the help of technology, ChatnLearn can enhance students’ oracy and eventually their learning outcomes.

Functionality
ChatnLearn offers two key learning modes:
1. Talk to Learn (Subject Learning) helps students build understanding by guiding them to give oral explanations on academic topics. By formulating instant responses, students connect ideas, identify gaps in their knowledge and gradually develop critical thinking.
2. Learn to Talk (Language Practice) focuses on helping students gain fluency and confidence when speaking in a second language such as English through regular conversational practice.

Each session includes pre-conversation notes and post-conversation feedback with suggestions for improvement and extra learning resources. This structure ensures students can see clear progress over time. Teachers and parents can also access these learning analytics and support students even more.

Feasibility
Using Large Language Models (LLMs) for personalized, oracy-based learning, technology is the cornerstone of ChatnLearn. Similar to Siri, our conversational AI combines Speech-to-Text, an LLM, Text-to-Speech, and logic for natural interaction. While technology exists, the solutions available in the market charge for high subscription fees. Therefore, we hope to build our own open-source solution to keep it economically sustainable. With conversational AI, ChatnLearn provides a safe, judgment-free sandbox for students to practice repeatedly and eventually build confidence in expressing their ideas through speech.

Evaluation
We will start with a technical feasibility study using lightweight AI setups with a small student group, which explores technical performance and user experience. It also helps us to validate the market demand. Based on feasibility findings, we will develop the MVP: a mobile-friendly web application enabling speech-to-speech conversations across key learning topics. It will ensure highest AI ethics safety standards, including robust content filtering. After development, we will conduct a pilot study with 10–20 secondary students. It can evaluate the effectiveness of engaging oracy with AI avatars.
Impact will be measured through engagement data (e.g., conversations completed), student self-assessments (oracy, confidence), and qualitative feedback on learning experiences.

N/A

ChatnLearn provides innovative solution through bringing conversational AI into education to improve learning experiences through a unique focus on oracy, which addresses the topic of (7) Reliable and Relevant Generative AI Application in Education.

Existing EdTech includes language apps like Duolingo, which is grammar/vocabulary through text, and general AI tutors like Khanmigo offering text-based subject guidance. While useful, they often neglect the effective use of spoken language and separate language skills from subject knowledge acquisition. The core gaps are therefore twofold: the lack of tools leveraging Oracy for deep learning across subjects, and the need for accessible, affordable personalised support.

Our creative solution is a voice-first conversational AI that integrates Oracy principles across two modes:
1) “Talk to Learn”: This is innovative in how it reframes learning. Students do not passively receive information; instead, they undertake active explaining aloud of subject concepts (e.g., Science) to an AI avatar. Articulation, central to Oracy, is used to push for improvement by elaborating thinking, ensuring active recall, and building understanding in the subject context.
3) “Learn to Talk”: This strengthens the above through targeted practice in spoken fluency and confidence, applying Oracy skills in a safe, non-judgmental “sandbox”. This is critical to help users cope with challenges like the fear of giving opinions in class and the lack of individualised attention.

Crucially, our innovation lies in using conversational AI to power a novel learning approach. Our “Talk to Learn” mode integrates speaking into subject understanding, while “Learn to Talk” builds essential fluency and confidence. This specific application of AI delivers personalized, effective learning support in a way that is significantly more affordable and accessible.

The core of ChatnLearn is its conversational AI interface. Once built, it can easily scale to cover a wide range of subjects. For "Talk to Learn", each conversation represents a learning topic. By designing guided dialogues for each subject, we can gradually cover curricula across areas like Science, Economics, and more. This flexible model allows expansion into new subjects without major technical changes.

For "Learn to Talk", the application can grow even further. While initially designed for English learning, the same framework can support other languages using GenAI. In the future, it can also help users prepare for interviews, public speaking, and presentations. This flexibility means ChatnLearn could easily enter new markets and adapt to other education systems globally if the model proves successful.

One foreseeable bottleneck is maintaining high-quality educational interactions. While the technology is feasible, it is also crucial to ensure every conversation meets learning goals. Strong, well-designed prompts for the AI are key. We plan to invite teachers and prompt engineers to refine prompts, ensuring conversations are meaningful and goal-driven.

Another rising concern is the responsible and ethical use of GenAI. As ChatnLearn targets secondary school students, we must ensure our learning environment is safe and trustworthy. Guidelines will be built into the AI’s design to prevent harmful content and ensure accuracy, overriding all other rules when necessary.

To ensure sustainability, ChatnLearn will nurture daily learning habits. We plan to introduce game-based elements like streaks and achievement badges to encourage usage for short but regular sessions. If chances allowed, ChatnLearn would intend to establish partnership with schools to provide e-learning platform services such as the www.prof-ho.com (每日一篇）to promote the concept of daily oral practices.

Social Impact
At ChatnLearn, we believe that quality education should be accessible by every learner, and so should quality technology. Our solution achieves equity and inclusion in two ways. First, we provide a second chance for passive learners in class to speak up, build confidence, and be heard. Second, ChatnLearn is an affordable alternative to private tutoring for disadvantaged students to access high-quality learning support. Through developing strong communication skills, students could become more competitive for future study and job opportunities, which eventually increases social mobility.

Responsibility
An inescapable responsibility of any EdTech is the ethical use of technology. At ChatnLearn, we take this seriously by ensuring that all content is safe, accurate, and age-appropriate for secondary school learners. Each conversation will be tested and safeguarded by our educator members to prevent harmful, biased, or inappropriate outputs. Clear guidelines will be embedded into the AI’s design to uphold fairness, honesty, and respect. It is our first principle that the learning environment we build must be trustworthy and promote responsible use of AI while supporting students’ growth.

Evolving Needs
Alongside product development, ChatnLearn will conduct longitudinal studies to track improvements in students’ speaking fluency, critical thinking, and confidence. The data collected will allow us to investigate whether integrating oracy with GenAI truly enhances learning outcomes. This could turn into a research project that contributes valuable academic insights into the role of conversational AI in education. ChatnLearn hopes to grow together with the evolving needs from the users. Just like our vision, we aim to support long-term educational innovation while ensuring that every learner’s voice is heard and empowered.

Personal Information Collection Statement (PICS):
1. The personal data collected in this form will be used for activity-organizing, record keeping and reporting only. The collected personal data will be purged within 6 years after the event.
2. Please note that it is obligatory to provide the personal data required.
3. Your personal data collected will be kept by the LTTC and will not be transferred to outside parties.
4. You have the right to request access to and correction of information held by us about you. If you wish to access or correct your personal data, please contact our staff at lttc@eduhk.hk.
5. The University’s Privacy Policy Statement can be access at https://www.eduhk.hk/en/privacy-policy.

• I have read and agree to the competition rules and privacy policy.