Emotive Translation Bubbles

K-pop live streams create a deeply engaging and emotional experience for fans worldwide. However, language barriers and imperfect machine translation (MT) models often lead to misinterpretations or loss of emotional nuance, making it difficult for non-Korean-speaking fans to fully connect with their favorite artists. While MT tools provide word-for-word translations, they fail to convey tone, sentiment, and emotional intent, which are crucial for understanding live conversations.

‍

Our goal is to improve content understanding for global audiences by supplementing machine translation with emotive translation bubbles—visual cues that represent the emotional tone of a conversation. By integrating expressively shaped speech bubbles into MT output, we aim to help fans better interpret emotions in live chat exchanges, creating a more immersive and emotionally connected viewing experience.

We designed expressive speech bubbles to accompany translated text, visually representing emotions such as joy, love, sadness, and mixed emotions. Our research involved informal user feedback sessions with K-pop fans, which showed that:

✔️ Fans enjoyed the added emotional context, finding it easier to grasp the overall sentiment.

✔️ Emotive bubbles helped users understand the gist of conversations without relying on direct word-for-word translation.

✔️ Some users felt a stronger emotional connection due to the enhanced expressiveness of the bubbles.