arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~

检索效果更好,鲁棒性更差:多跳检索增强生成(RAG)如何放大上游自动语音识别(ASR)错误

Better Retrieval, Worse Robustness: How Multi-hop RAG Amplifies Upstream ASR Errors

Zhenghua Bao

arXiv 2608.22872首次发表:更新:

发表机构

Continuum AI(Continuum AI)

机构由 AI 辅助整理,请以论文原文为准。

AI 中文总结

该研究探究多跳RAG的实体图链接、迭代重述两种扩展方式对上游ASR错误的影响,发现二者会放大错误,其组合使F1差距增大36%-67%,实体损坏是主要失效模式,轻量缓解措施效果有限。

AI 中文摘要

基于语音的应用会在任何检索模块之前,通过自动语音识别(ASR)处理语音查询,因此ASR错误会作为固定的上游约束进入整个流程。我们通过实证测试了标准检索增强生成(RAG)的两种扩展方式——实体图链接与迭代重述,是会吸收还是放大这些错误。我们使用神经文本转语音(TTS)合成的四种英语口音,在三个多跳问答基准(HotpotQA、2WikiMultiHopQA和MuSiQue)上,针对纯文本 oracle 评估了四种RAG配置。尽管结构更丰富的配置在ASR输入下通常保持更高的绝对F1值,但两种扩展方式都会放大错误:在所有三个基准上,它们的组合下,从纯文本到最高词错误率(WER)口音的F1差距,比在朴素密集检索下大36%-67%。主要失效模式是一个或多个查询实体被损坏,在所有四种方法中,2WikiMultiHopQA上87%-96%的退化案例都由该原因导致。两种轻量的表面形式缓解措施几乎未缩小差距,表明下游检索结构会放大残留的实体错误。我们在该httpsURL发布代码与数据。

英文摘要

Speech-based applications pass spoken queries through automatic speech recognition (ASR) before any retrieval module, so ASR errors enter the pipeline as a fixed upstream constraint. We empirically test whether two extensions to standard retrieval-augmented generation (RAG), entity-graph linking and iterative reformulation, absorb or amplify these errors. Using four English accents synthesized through neural TTS, we evaluate four RAG configurations on three multi-hop QA benchmarks (HotpotQA, 2WikiMultiHopQA and MuSiQue) against a clean-text oracle. Although the structurally richer configurations generally retain higher absolute F1 under ASR input, both extensions amplify the error: the F1 gap from clean text to the highest-WER accent is 36-67% larger under their combination than under naive dense retrieval, on all three benchmarks. The dominant failure mode is corruption of one or more query entities, accounting for 87-96% of degradation cases on 2WikiMultiHopQA across all four methods. Two lightweight surface-form mitigations leave most of the gap intact, indicating that downstream retrieval structure amplifies remaining entity errors. We release code and data at https://github.com/Continuum-AI-Corp/spoken-multihop-rag .

CommentsAccepted to EMNLP 2026 (Main Conference)

论文原文

arXiv 摘要页 · PDF 原文 · HTML 原文

↑