Speech LLM-based ASR often struggles with named entities and long-tail words due to strong internal language-model priors. Retrieval-augmented biasing can help, but its effectiveness depends on accurate hotword localization in full-utterance speech under weak supervision. We propose CLAR, a dual-encoder speech-text retriever that uses Continuous Integrate-and-Fire (CIF) to learn monotonic token-level alignments without timestamps. With length-aware localized matching, CLAR anchors short-entity acoustic cues and reduces representation dilution and attention drift. The retriever is trained with a multi-granularity objective combining global and local segment-level contrastive losses and a CIF quantity constraint. At inference, top-ranked hotwords are injected as contextual prompts for the Speech LLM, improving recognition without shallow fusion. Experiments show that CLAR significantly improves hotword retrieval and reduces both CER and B-WER against strong contextual ASR baselines.

翻译：基于语音大语言模型的自动语音识别（ASR）常因强内部语言模型先验而难以处理命名实体和长尾词汇。检索增强偏置可提供帮助，但其效果取决于在弱监督条件下对全句语音中热词的准确定位。本文提出CLAR——一种采用连续积分-触发（CIF）机制学习单调令牌级对齐（无需时间戳）的双编码器语音-文本检索器。通过长度感知的局部匹配，CLAR锚定短实体声学线索，减少表征稀释和注意力漂移。该检索器采用多粒度目标进行训练，结合全局与局部片段级对比损失及CIF数量约束。推理时，将排名靠前的热词作为上下文提示注入语音大语言模型，无需浅融合即可提升识别效果。实验表明，CLAR显著提升了热词检索性能，并在强上下文ASR基线基础上降低了词错误率（CER）和短语级词错误率（B-WER）。