The False Resonance: A Critical Examination of Emotion Embedding Similarity for Speech Generation Evaluation
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911632177758208 |
|---|---|
| author | Tsai, Yun-Shao Lin, Yi-Cheng Chou, Huang-Cheng Hsu, Tzu-Wen Hsu, Yun-Man Chen, Chun Wei Narayanan, Shrikanth Lee, Hung-yi |
| author_facet | Tsai, Yun-Shao Lin, Yi-Cheng Chou, Huang-Cheng Hsu, Tzu-Wen Hsu, Yun-Man Chen, Chun Wei Narayanan, Shrikanth Lee, Hung-yi |
| contents | Objective metrics for emotional expressiveness are vital for speech generation, particularly in expressive synthesis and voice conversion requiring emotional prosody transfer. To quantify this, the field widely relies on emotion similarity between reference and generated samples. This approach computes cosine similarity of embeddings from encoders like emotion2vec, assuming they capture affective cues despite linguistic and speaker variations. We challenge this assumption through controlled adversarial tasks and human alignment tests. Despite high classification accuracy, these latent spaces are unsuitable for zero-shot similarity evaluation. Representational limitations cause linguistic and speaker interference to overshadow emotional features, degrading discriminative ability. Consequently, the metric misaligns with human perception. This acoustic vulnerability reveals it rewards acoustic mimicry over genuine emotional synthesis. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_26347 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | The False Resonance: A Critical Examination of Emotion Embedding Similarity for Speech Generation Evaluation Tsai, Yun-Shao Lin, Yi-Cheng Chou, Huang-Cheng Hsu, Tzu-Wen Hsu, Yun-Man Chen, Chun Wei Narayanan, Shrikanth Lee, Hung-yi Audio and Speech Processing Computation and Language Objective metrics for emotional expressiveness are vital for speech generation, particularly in expressive synthesis and voice conversion requiring emotional prosody transfer. To quantify this, the field widely relies on emotion similarity between reference and generated samples. This approach computes cosine similarity of embeddings from encoders like emotion2vec, assuming they capture affective cues despite linguistic and speaker variations. We challenge this assumption through controlled adversarial tasks and human alignment tests. Despite high classification accuracy, these latent spaces are unsuitable for zero-shot similarity evaluation. Representational limitations cause linguistic and speaker interference to overshadow emotional features, degrading discriminative ability. Consequently, the metric misaligns with human perception. This acoustic vulnerability reveals it rewards acoustic mimicry over genuine emotional synthesis. |
| title | The False Resonance: A Critical Examination of Emotion Embedding Similarity for Speech Generation Evaluation |
| topic | Audio and Speech Processing Computation and Language |
| url | https://arxiv.org/abs/2604.26347 |