aiqbench analysis
This research introduces ER-EDF, a framework that decouples emotion perception from regulation within large audio-language models to move beyond simple affect mirroring toward calibrated empathetic support. By grounding the approach in psychological theories like the Perception-Action Model, it demonstrates improved response quality across multiple LALMs using new empathy-aware evaluation metrics. The work highlights a critical architectural shift for spoken dialogue systems, emphasizing that effective AI empathy requires explicit regulatory mechanisms rather than direct conditioning on emotional signals.
ER-EDF: A Psychology-Grounded Emotion Regulation Framework for Speech Empathetic Dialogue Generation in Large Audio-Language Models
arXiv:2609.15089v1 Announce Type: new Abstract: Empathetic response generation in spoken dialogue systems requires both accurate emotion perception and appropriate emotion regulation. Grounded in psychological theories such as the Perception-Action Model and emotion regulation theory, effective empathy depends not only on inferring a user's affective state but also on regulating how it is expressed in responses. However, recent large audio-language models (LALMs) largely treat emotion as a direct conditioning signal, lacking explicit regulatory mechanisms, which often leads to affect mirroring rather than calibrated support. We propose ER-EDF, a psychology-grounded framework that explicitly decouples emotion perception and emotion regulation in LALMs. Perception tracks the user's emotional state, while regulation determines how this state should guide empathetic response generation. The framework is model-agnostic and integrates seamlessly into existing LALMs. We further construct a spoken empathetic dialogue dataset and introduce empathy-aware evaluation metrics beyond lexical matching. Experiments across five LALMs and two datasets show that ER-EDF consistently improves empathetic response quality in both automatic and human evaluations, highlighting the importance of jointly modeling emotion perception and regulation in spoken empathetic dialogue systems, paving a new direction for psychologically grounded empathetic AI.
Abstract:Empathetic response generation in spoken dialogue systems requires both accurate emotion perception and appropriate emotion regulation. Grounded in psychological theories such as the Perception-Action Model and emotion regulation theory, effective empathy depends not only on inferring a user's affective state but also on regulating how it is expressed in responses. However, recent large audio-language models (LALMs) largely treat emotion as a direct conditioning signal, lacking explicit regulatory mechanisms, which often leads to affect mirroring rather than calibrated support. We propose ER-EDF, a psychology-grounded framework that explicitly decouples emotion perception and emotion regulation in LALMs. Perception tracks the user's emotional state, while regulation determines how this state should guide empathetic response generation. The framework is model-agnostic and integrates seamlessly into existing LALMs. We further construct a spoken empathetic dialogue dataset and introduce empathy-aware evaluation metrics beyond lexical matching. Experiments across five LALMs and two datasets show that ER-EDF consistently improves empathetic response quality in both automatic and human evaluations, highlighting the importance of jointly modeling emotion perception and regulation in spoken empathetic dialogue systems, paving a new direction for psychologically grounded empathetic AI.
Artificial Intelligence (cs.AI)
arXiv:2609.15089 [cs.AI]
(or arXiv:2609.15089v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2609.15089 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 276 words
Related coverage
Models Can't Remember Their Training. Neither Can You.
LessWrong · Jul 19, 2026
Multimodal Detection of Higher-Order Behavioral Constructs: Self-Compassion in Structured Reflective Interaction
arXiv · Oct 1, 2026
When Text Misleads: Inconsistent-Aware Reasoning for Audio-Grounded Dialogue
arXiv · Aug 28, 2026
CompanionBench: A Theory-Anchored, Real-World-Grounded Benchmark for AI Emotional Companionship
arXiv · Aug 5, 2026