<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="hsu-etal-2025-lets">
    <titleInfo>
        <title>Let’s Fuse Step by Step: A Generative Fusion Decoding Algorithm with LLMs for Robust and Instruction-Aware ASR and OCR</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Chan-Jan</namePart>
        <namePart type="family">Hsu</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yi-Chang</namePart>
        <namePart type="family">Chen</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Feng-Ting</namePart>
        <namePart type="family">Liao</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Pei-Chen</namePart>
        <namePart type="family">Ho</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yu-Hsiang</namePart>
        <namePart type="family">Wang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Po-Chun</namePart>
        <namePart type="family">Hsu</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Da-shan</namePart>
        <namePart type="family">Shiu</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2025-07</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Findings of the Association for Computational Linguistics: ACL 2025</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Wanxiang</namePart>
            <namePart type="family">Che</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Joyce</namePart>
            <namePart type="family">Nabende</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Ekaterina</namePart>
            <namePart type="family">Shutova</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Mohammad</namePart>
            <namePart type="given">Taher</namePart>
            <namePart type="family">Pilehvar</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Vienna, Austria</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-256-5</identifier>
    </relatedItem>
    <abstract>We introduce “Generative Fusion Decoding” (GFD), a novel shallow fusion framework, utilized to integrate large language models(LLMs) into cross-modal text recognition systems inlculding automatic speech recognition (ASR) and optical character recognition (OCR). We derive the formulas necessary to enable GFD to operate across mismatched token spaces of different models by calculating likelihood at the byte level, thereby enabling seamless fusion and synchronous progression during the decoding process. GFD is plug-and-play bydesign, making it readily compatible with various auto-regressive models without the need for any re-training. GFD proves effective for general ASR and OCR tasks through intermediate and frequent interactions with LLMs, surpassing cascaded methods in English and Mandarin benchmarks. In addition, GFD transfers in-context learning abilities of LLMs and allows for adaptive ASR in instruction-aware andlong-context settings, yielding significant WER reductions of up to 17.7%.</abstract>
    <identifier type="citekey">hsu-etal-2025-lets</identifier>
    <identifier type="doi">10.18653/v1/2025.findings-acl.1281</identifier>
    <location>
        <url>https://aclanthology.org/2025.findings-acl.1281/</url>
    </location>
    <part>
        <date>2025-07</date>
        <extent unit="page">
            <start>24959</start>
            <end>24973</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Let’s Fuse Step by Step: A Generative Fusion Decoding Algorithm with LLMs for Robust and Instruction-Aware ASR and OCR
%A Hsu, Chan-Jan
%A Chen, Yi-Chang
%A Liao, Feng-Ting
%A Ho, Pei-Chen
%A Wang, Yu-Hsiang
%A Hsu, Po-Chun
%A Shiu, Da-shan
%Y Che, Wanxiang
%Y Nabende, Joyce
%Y Shutova, Ekaterina
%Y Pilehvar, Mohammad Taher
%S Findings of the Association for Computational Linguistics: ACL 2025
%D 2025
%8 July
%I Association for Computational Linguistics
%C Vienna, Austria
%@ 979-8-89176-256-5
%F hsu-etal-2025-lets
%X We introduce “Generative Fusion Decoding” (GFD), a novel shallow fusion framework, utilized to integrate large language models(LLMs) into cross-modal text recognition systems inlculding automatic speech recognition (ASR) and optical character recognition (OCR). We derive the formulas necessary to enable GFD to operate across mismatched token spaces of different models by calculating likelihood at the byte level, thereby enabling seamless fusion and synchronous progression during the decoding process. GFD is plug-and-play bydesign, making it readily compatible with various auto-regressive models without the need for any re-training. GFD proves effective for general ASR and OCR tasks through intermediate and frequent interactions with LLMs, surpassing cascaded methods in English and Mandarin benchmarks. In addition, GFD transfers in-context learning abilities of LLMs and allows for adaptive ASR in instruction-aware andlong-context settings, yielding significant WER reductions of up to 17.7%.
%R 10.18653/v1/2025.findings-acl.1281
%U https://aclanthology.org/2025.findings-acl.1281/
%U https://doi.org/10.18653/v1/2025.findings-acl.1281
%P 24959-24973

Markdown (Informal)

[Let’s Fuse Step by Step: A Generative Fusion Decoding Algorithm with LLMs for Robust and Instruction-Aware ASR and OCR](https://aclanthology.org/2025.findings-acl.1281/) (Hsu et al., Findings 2025)

• Let’s Fuse Step by Step: A Generative Fusion Decoding Algorithm with LLMs for Robust and Instruction-Aware ASR and OCR (Hsu et al., Findings 2025)

ACL

• Chan-Jan Hsu, Yi-Chang Chen, Feng-Ting Liao, Pei-Chen Ho, Yu-Hsiang Wang, Po-Chun Hsu, and Da-shan Shiu. 2025. Let’s Fuse Step by Step: A Generative Fusion Decoding Algorithm with LLMs for Robust and Instruction-Aware ASR and OCR. In Findings of the Association for Computational Linguistics: ACL 2025, pages 24959–24973, Vienna, Austria. Association for Computational Linguistics.