Salvato in:
Dettagli Bibliografici
Autori principali: Xue, Xiangyuan, Lu, Jiajun, Gao, Yan, Huang, Gongping, Dang, Ting, Jia, Hong
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:https://arxiv.org/abs/2603.11397
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866912962895151104
author Xue, Xiangyuan
Lu, Jiajun
Gao, Yan
Huang, Gongping
Dang, Ting
Jia, Hong
author_facet Xue, Xiangyuan
Lu, Jiajun
Gao, Yan
Huang, Gongping
Dang, Ting
Jia, Hong
contents Speech Emotion Captioning (SEC) leverages large audio-language models to generate rich, context-aware affective descriptions from speech. However, real-world deployment remains challenging due to the substantial computational demands on resource-constrained edge devices and the privacy risks of transmitting biometric audio. While smaller audio-language models enable efficient on-device SEC, their limited capacity often weakens subtle paralinguistic modeling and fine-grained affective grounding. We propose an edge-cloud collaborative framework based on Uncertainty-Guided Speculative Decoding (UGSD). A lightweight edge model drafts captions locally, and only high-uncertainty token blocks are selectively escalated to a stronger cloud verifier for validation. Experiments on the MER2024 benchmark demonstrate substantial BLEU improvements up to 62.7%. UGSD further achieves 1.4x lower latency and 8.5x higher token throughput compared to an edge-only model. These results empirically characterize the quality-efficiency-privacy trade-off in deployable SEC systems.
format Preprint
id arxiv_https___arxiv_org_abs_2603_11397
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Edge-Cloud Collaborative Speech Emotion Captioning via Token-Level Speculative Decoding in Audio-Language Models
Xue, Xiangyuan
Lu, Jiajun
Gao, Yan
Huang, Gongping
Dang, Ting
Jia, Hong
Sound
Speech Emotion Captioning (SEC) leverages large audio-language models to generate rich, context-aware affective descriptions from speech. However, real-world deployment remains challenging due to the substantial computational demands on resource-constrained edge devices and the privacy risks of transmitting biometric audio. While smaller audio-language models enable efficient on-device SEC, their limited capacity often weakens subtle paralinguistic modeling and fine-grained affective grounding. We propose an edge-cloud collaborative framework based on Uncertainty-Guided Speculative Decoding (UGSD). A lightweight edge model drafts captions locally, and only high-uncertainty token blocks are selectively escalated to a stronger cloud verifier for validation. Experiments on the MER2024 benchmark demonstrate substantial BLEU improvements up to 62.7%. UGSD further achieves 1.4x lower latency and 8.5x higher token throughput compared to an edge-only model. These results empirically characterize the quality-efficiency-privacy trade-off in deployable SEC systems.
title Edge-Cloud Collaborative Speech Emotion Captioning via Token-Level Speculative Decoding in Audio-Language Models
topic Sound
url https://arxiv.org/abs/2603.11397