MoVE: Translating Laughter and Tears via Mixture of Vocalization Experts in Speech-to-Speech Translation
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917420576276480 |
|---|---|
| author | Chen, Szu-Chi Tsai, I-Ning Lin, Yi-Cheng Huang, Sung-Feng Lee, Hung-yi |
| author_facet | Chen, Szu-Chi Tsai, I-Ning Lin, Yi-Cheng Huang, Sung-Feng Lee, Hung-yi |
| contents | Recent Speech-to-Speech Translation (S2ST) systems achieve strong semantic accuracy yet consistently strip away non-verbal vocalizations (NVs), such as laughter and crying that convey pragmatic intent, which severely limits real-world utility. We address this via three contributions. First, we propose a synthesis pipeline for building scalable expressive datasets to overcome the data scarcity limitation. Second, we propose MoVE, a Mixture-of-LoRA-Experts architecture with expressive-specialized adapters and a soft-weighting router that blends experts for capturing hybrid expressive states. Third, we show pretrained AudioLLMs enable striking data efficiency: 30 minutes of curated data is enough for strong performance. On English-Chinese S2ST, while comparing with strong baselines, MoVE reproduces target NVs in 76% of cases and achieves the highest human-rated naturalness and emotional fidelity among all compared systems, where existing S2ST systems preserve at most 14% of NVs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_17435 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | MoVE: Translating Laughter and Tears via Mixture of Vocalization Experts in Speech-to-Speech Translation Chen, Szu-Chi Tsai, I-Ning Lin, Yi-Cheng Huang, Sung-Feng Lee, Hung-yi Computation and Language Artificial Intelligence Sound Audio and Speech Processing Recent Speech-to-Speech Translation (S2ST) systems achieve strong semantic accuracy yet consistently strip away non-verbal vocalizations (NVs), such as laughter and crying that convey pragmatic intent, which severely limits real-world utility. We address this via three contributions. First, we propose a synthesis pipeline for building scalable expressive datasets to overcome the data scarcity limitation. Second, we propose MoVE, a Mixture-of-LoRA-Experts architecture with expressive-specialized adapters and a soft-weighting router that blends experts for capturing hybrid expressive states. Third, we show pretrained AudioLLMs enable striking data efficiency: 30 minutes of curated data is enough for strong performance. On English-Chinese S2ST, while comparing with strong baselines, MoVE reproduces target NVs in 76% of cases and achieves the highest human-rated naturalness and emotional fidelity among all compared systems, where existing S2ST systems preserve at most 14% of NVs. |
| title | MoVE: Translating Laughter and Tears via Mixture of Vocalization Experts in Speech-to-Speech Translation |
| topic | Computation and Language Artificial Intelligence Sound Audio and Speech Processing |
| url | https://arxiv.org/abs/2604.17435 |