Research
Exploring the frontiers of technology through academic research and innovation
Keywords
EmoTale: Shaping Narrative Trajectories with Reader's Affective States
Authors: B. Susindra Reddy (Lead), K. Sai Manogyna, M. Vasudevan, et al.
Published in: Challenges in Information, Communication and Computing Technology – Vol. 2 (2024) Nov
In book: Challenges in Information, Communication and Computing Technology (pp.678-683)
Summary
EmoTale is an intelligent storytelling system that generates dynamic, emotionally personalized narratives using reader feedback. It leverages multimodal emotion recognition (via voice tone and facial cues) and integrates this feedback into a custom-built large language model (LLM) pipeline to shape story trajectories in real time.
The system offers a two-part narrative experience:
- Users provide a moral or theme.
- The model adapts the story at each stage based on their detected emotional responses (joy, fear, interest, etc.).
Its modular architecture includes:
- Emotion Detection module (voice/facial input)
- LLM-based Story Generator (custom trained)
- Adaptive Response Engine
- UI + Persistence layer for continuity
System Architecture
My Role (Lead Developer & Researcher)
- Led the entire research and development process.
- Built and fine-tuned the LLM-based narrative engine.
- Designed the emotion feedback loop, integrating real-time user emotion into story shaping.
- Supervised teammates and guided model integration, UI design, and deployment strategy.
- Deployed the full-stack system, ensuring real-time performance and personalization.