MemeSpeak: Making Image Memes Blind-Accessible through Emotional Speech Augmentation

Image memes—combining humorous or satirical pictures with overlaying text—are a powerful medium for emotional expression. However, their visual nature makes them largely inaccessible to blind and low vision (BLV) users. We present MemeSpeak, a system that generates emotionally expressive audio narrations of memes through emotional speech augmentation. We developed two emotion-conveying strategies: emotion-only and character-plus-emotion, and evaluated them against standard screen-reader-based descriptions with 12 BLV participants. Results showed that both strategies significantly improved emotional resonance, eliciting stronger emotional fluctuations and better understanding of the sender’s intent. Qualitative feedback revealed that the character-plus-emotion strategy enhanced immersion and clarity, particularly for complex or multi-character memes. We further highlight implications for conveying visual humor in blind-accessible formats.