On the Emotion Understanding of Synthesized Speech

TL;DR

This paper critically evaluates whether current speech emotion recognition models can accurately understand emotion in synthesized speech, revealing significant limitations due to representation mismatch and reliance on textual semantics.

Contribution

It systematically assesses SER models on synthesized speech, highlighting their inability to generalize and exposing the challenges in capturing paralinguistic cues.

Findings

SER models do not generalize well to synthesized speech

Representation mismatch caused by speech token prediction affects emotion recognition

Generative SLMs rely more on textual semantics than paralinguistic cues

Abstract

Emotion is a core paralinguistic feature in voice interaction. It is widely believed that emotion understanding models learn fundamental representations that transfer to synthesized speech, making emotion understanding results a plausible reward or evaluation metric for assessing emotional expressiveness in speech synthesis. In this work, we critically examine this assumption by systematically evaluating Speech Emotion Recognition (SER) on synthesized speech across datasets, discriminative and generative SER models, and diverse synthesis models. We find that current SER models can not generalize to synthesized speech, largely because speech token prediction during synthesis induces a representation mismatch between synthesized and human speech. Moreover, generative Speech Language Models (SLMs) tend to infer emotion from textual semantics while ignoring paralinguistic cues. Overall, our findings suggest that existing SER models often exploit non-robust shortcuts rather than capturing fundamental features, and paralinguistic understanding in SLMs remains challenging.