TY  - RPRT
TI  - Edge-Cloud Collaborative Speech Emotion Captioning via Token-Level Speculative Decoding in Audio-Language Models
AU  - Xiangyuan Xue
AU  - Jiajun Lu
AU  - Yan Gao
AU  - Gongping Huang
AU  - Ting Dang
AU  - Hong Jia
PY  - 2026
UR  - https://arxiv.org/abs/2603.11397
ID  - 2603.11397
ER  -