arXiv cs.CL· Elahe Khatibi, Ziyu Wang, Amir M. Rahmani·· 11 小时前AI 评分31
MedAstra：无需 ASR 的语音检索框架，为医疗问答提供诊断依据
MedAstra: ASR-less Speech Retrieval with Diagnostic Grounding for Medical QA
AI 导读
MedAstra 提出一种无需 ASR 的语音检索框架，将冻结的 WavLM 语音表示经轻量适配器映射进冻结的 E5 文本嵌入空间，可直接以波形作为查询。语音检索器 Recall@10 达 87.6，混合配置达 90.4，高于转写优先稠密检索的 81.0。检索到的段落与语音关联文本元数据用于生成带实体替代项和诊断对比的答案，但词法检索与文本生成仍需文本输入。
正文 · 原文
Abstract:Spoken medical question answering can fail when automatic speech recognition (ASR) corrupts a clinically important term before retrieval. We present MedAstra, a framework that maps frozen WavLM speech representations through a lightweight adapter into a frozen E5 text embedding space. This enables a dense retrieval path that accepts a waveform without a query transcript. Retrieved passages and speech-linked textual metadata support answer generation with entity alternatives, diagnostic contrasts, relation heuristics, and bounded retrieval refinement. ASR-less refers to the initial speech retrieval interface; lexical retrieval, string-based entity correction, and the text generator require textual inputs. The deterministic speech embedding does not constitute a calibrated model of acoustic uncertainty. Reported retrieval results reach 87.6 Recall@10 for the speech-only retriever and 90.4 for the hybrid configuration, compared with 81.0 for transcript-first dense retrieval. Medical QA and ablations provide descriptive system comparisons, with input-provenance and evaluation limitations stated explicitly. We distinguish controlled synthetic acoustic tests from real clinical speech, qualify component attribution, and specify the remaining comparisons needed to assess speech-only QA and clinical generalization.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2611.01261 [cs.CL] |
| (or arXiv:2611.01261v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2611.01261 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Elahe Khatibi [view email]
[v1]
Sat, 31 Oct 2026 21:00:22 UTC (10,403 KB)
来源：arXiv cs.CL · arxiv.org