EMO-TTA: Improving Test-Time Adaptation of Audio-Language Models for Speech Emotion Recognition
This work addresses the problem of test-time adaptation for speech emotion recognition, offering a practical solution for out-of-domain scenarios, though it appears incremental as it builds on existing TTA methods.
The paper tackled performance degradation in speech emotion recognition due to distribution shifts at test time by proposing Emo-TTA, a lightweight, training-free adaptation framework that updates class-conditional statistics via Expectation-Maximization, resulting in consistent accuracy improvements on six out-of-domain benchmarks.
Speech emotion recognition (SER) with audio-language models (ALMs) remains vulnerable to distribution shifts at test time, leading to performance degradation in out-of-domain scenarios. Test-time adaptation (TTA) provides a promising solution but often relies on gradient-based updates or prompt tuning, limiting flexibility and practicality. We propose Emo-TTA, a lightweight, training-free adaptation framework that incrementally updates class-conditional statistics via an Expectation-Maximization procedure for explicit test-time distribution estimation, using ALM predictions as priors. Emo-TTA operates on individual test samples without modifying model weights. Experiments on six out-of-domain SER benchmarks show consistent accuracy improvements over prior TTA baselines, demonstrating the effectiveness of statistical adaptation in aligning model predictions with evolving test distributions.