Saved in:
Bibliographic Details
Main Authors: Huang, Shangkun, Shen, Huan, Zou, Wei, Chen, Yunzhang
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.25460
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912983798513664
author Huang, Shangkun
Shen, Huan
Zou, Wei
Chen, Yunzhang
author_facet Huang, Shangkun
Shen, Huan
Zou, Wei
Chen, Yunzhang
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2603_25460
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle CLAR: CIF-Localized Alignment for Retrieval-Augmented Speech LLM-Based Contextual ASR
Huang, Shangkun
Shen, Huan
Zou, Wei
Chen, Yunzhang
Sound
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.
title CLAR: CIF-Localized Alignment for Retrieval-Augmented Speech LLM-Based Contextual ASR
topic Sound
url https://arxiv.org/abs/2603.25460