Shannon Brownlee - Emotion Models for Large-Scale Voice Interaction

As text-based generative AI systems have advanced rapidly, attention is shifting to the next frontier: voice. Unlike text, voice carries dense paralinguistic signals like emotional cues that are often lost in traditional NLP systems, contributing to the uncanny valley in voice agents. This talk will dive into the research and architecture behind Valence’s approach to real-time emotion detection at scale, where deep learning models detect emotional shifts across diverse speakers and environments. Shannon will outline the data pipelines, model training considerations, and multi-modal validation strategies that make high-fidelity emotion detection feasible, and discuss open research challenges in scaling emotion AI for production systems.