Visualizing speech styles in captions for deaf and hard-of-hearing viewers

Speech styles such as extension, emphasis, and pause play an important role in capturing the audience's attention and conveying a message accurately. Unfortunately, it is challenging for Deaf and Hard-of-Hearing (DHH) people to enjoy these benefits when watching lectures with common captions. In this paper, we propose a new caption system that automatically analyzes speech styles from audio and visualizes them using visualization elements such as punctuation, paint-on, color, and boldness. We conducted a comparative study with 26 DHH viewers and found that the proposed caption system enabled them to recognize the speaker's speech style in lectures. As a result, the DHH viewers were able to watch lecture videos more vividly and were more engaged with the lectures. In particular, punctuation can be a practical solution to visualize speech styles and ensure legibility. Participants expressed a desire to use our caption system in their daily lives, providing valuable insights for future sound-visualized caption research.