EvolveCaptions: Empowering DHH Users Through Real-Time Collaborative Captioning
2025-10-03
Summary
EvolveCaptions is a system designed to improve accessibility in conversations involving Deaf and Hard of Hearing (DHH) individuals by enhancing automatic speech recognition (ASR) accuracy through real-time, collaborative corrections. The system allows hearing participants to correct live captions during conversations, and these corrections are used to generate targeted prompts for DHH speakers to record, enabling the ASR model to adapt to their unique speech patterns over time. A study involving 12 DHH and 6 hearing participants showed that EvolveCaptions significantly reduced word error rates with minimal effort, suggesting its potential for more equitable communication.
Why This Matters
This article highlights a novel approach to improving communication accessibility for DHH individuals, who often face challenges with existing ASR systems. By focusing on real-time, collaborative adaptation, EvolveCaptions not only enhances speech recognition accuracy but also distributes the labor of adaptation across all participants, promoting a more inclusive and equitable communication environment. This is crucial for enabling meaningful participation of DHH individuals in educational, professional, and social settings.
How You Can Use This Info
Professionals working with or supporting DHH individuals can leverage systems like EvolveCaptions to facilitate more accessible and effective communication. Incorporating such technology into workplaces or educational environments can improve engagement and reduce communication barriers. Additionally, understanding and advocating for collaborative and adaptive technologies can help organizations meet accessibility standards and support diversity and inclusion initiatives. Interested parties can explore the EvolveCaptions GitHub repository for more technical details and potential integration.