Tatsuya Kawahara

Emotion preference models can be dramatically improved by addressing both data sparsity and model bias, leading to more accurate emotional assessments in multimodal contexts.

User satisfaction in conversational AI hinges more on natural memory integration than on traditional recall accuracy, revealing a critical gap in current evaluation methods.

Enhancing AI clones with listening behaviors can dramatically elevate user perceptions of authenticity and engagement.

Empathetic robots that understand and respond to emotional dynamics can significantly enhance user satisfaction and emotional alignment during interactions.

Relationships with conversational AIs evolve through both gradual accumulation of familiarity and sudden relational turning points that can be anticipated through user behavior.

Performance improvements in speech model fine-tuning are often an illusion, heavily reliant on the specific pretrained instance rather than true methodological advancements.

Real-time listener nodding can be generated with context-aware kinematic predictions, dramatically improving avatar interactions.

Leveraging spatial audio cues, PATSE achieves superior speaker extraction in chaotic multi-party conversations without the need for complex diarization.

Current LLMs may produce varied validating responses, but they fundamentally lack true emotional understanding, revealing a significant gap in dialogue system performance.

You can predict the best moment to offer emotional support just by listening to someone's voice, no text needed.