Revolutionizing Literacy in Indonesia through Innovative Technology and Localized Content

• In Event: Leveraging Voice AI (Artificial Intelligence) to assess early grade reading skills and support learning

Proposal

Indonesia's literacy landscape has long been a cause for concern. In a global literacy study, the nation ranked 60th out of 61 countries, highlighting a pressing need for innovative reading solutions. We stepped up to this challenge by developing a culturally relevant and advanced learning platform to boost literacy rates among Indonesian children.

Our journey began with an in-depth exploration of Indonesia's literacy landscape. We discovered that low reading interest stemmed partly from poor book quality and content that failed to captivate young readers. We engaged directly with students, teachers, and parents through surveys and interviews to address this. Their insights guided us in creating a series of localized storybooks that resonated with Indonesian children's interests and cultural context.

At the heart of our solution is our learn-to-read app, which seamlessly integrates these custom-designed books into an interactive learning experience. The app's standout feature is its real-time pronunciation feedback system, which gently corrects children's reading errors on the spot. This functionality is powered by our open-source Indonesian phoneme-level speech recognition technology, which has been tailored and developed specifically for this project.

We also created a prosodic children's text-to-speech system that runs efficiently on-device and offline, even on lightweight hardware. This approach ensures widespread accessibility across Indonesia's diverse technological landscape, where low-end devices are widely used.

Offline functionality is another key feature of our app, allowing access to its book library without an internet connection. Our speech recognition technology works offline, utilizing the device's processor and adapting to the unique challenges of children's speech patterns.

The app's design prioritizes engagement and retention, featuring locally inspired imagery that enhances the cultural relevance of the content. We've further improved the user experience with features like word syllabification, animated instructions, and natural voice synthesis, making learning educational and entertaining.

Our strategy has proven effective, with user data showing high retention rates and consistent improvement in reading skills. An efficacy study in reading fluency among children showed significant improvements.

For educators and parents, we offer a companion classroom app that tracks student progress. This tool provides valuable insights into each child's reading habits, preferences, and areas for improvement, enabling personalized instruction and support.

By combining deep cultural understanding with cutting-edge technology, we have created a solution that's effective, engaging, and accessible. The significant improvements in reading fluency and user engagement suggest that our app has the potential to make a substantial impact to resolve the learning crisis. We are currently developing our app in Kiswahili language. The common and emerging challenges associated with adapting this technology to additional low-resource languages will be discussed at this session.

Author

• Adrian DeWitts, Bookbot