论文研究
arxiv.org ·
超越提示：通过 Logit 空间集成（LOGIC）为语音大语言模型实现高效且稳健的上下文偏置
该论文提出 LOGIC（Logit-Space Integration for Contextual Biasing），在解码层直接进行上下文偏置，将上下文注入与输入处理解耦，使复杂度相对提示长度保持常数，以规避 prompting 的上下文窗口限制、推理延迟上升和 lost-in-the-middle 问题，并避免 GEC 后处理常见的过度纠正与未说实体幻觉。作者在 Phi-4-MM 上跨 11 个多语言场景实验，报告 Entity WER 平均相对下降 9%，False Alarm Rate 仅增加 0.30%。
AI推荐理由 arXiv 论文（cs.AI），作者为 Peidong Wang、Jian Xue、Jinyu Li，摘要明确给出方法与量化结果：LOGIC 在解码层做 logit 空间集成，将上下文注入与输入处理解耦，相对 prompt 长度保持常数复杂度；在 Phi-4-MM 上覆盖 11 个多语言场景，Entity WER 相对降低约 9%，False Alarm Rate 仅增加 0.30%。新信息在于提出替代 prompting 与 GEC 的第三种上下文偏置路径，并针对 prompting 的上下文窗口、延迟、lost-in-the-middle 以及 GEC 的过度纠正/实体幻觉问题给出方案。局限：仅摘要级证据，未提供数据集规模、基线细节与消融，效果为单一模型上的报告值，需正文验证。
78
正文
Computer Science > Artificial Intelligence
arXiv:2601.15397 (cs)
[Submitted on 21 Jan 2026 (v1), last revised 10 Sep 2026 (this version, v3)]
Title:Beyond Prompting: Efficient and Robust Contextual Biasing for Speech LLMs via Logit-Space Integration (LOGIC)
Authors:Peidong Wang, Jian Xue, Jinyu Li
View a PDF of the paper titled Beyond Prompting: Efficient and Robust Contextual Biasing for Speech LLMs via Logit-Space Integration (LOGIC), by Peidong Wang and 2 other authors
View PDF
HTML (experimental)
Abstract:The rapid emergence of new entities -- driven by cultural shifts, evolving trends, and personalized user data -- poses a significant challenge for existing Speech Large Language Models (Speech LLMs). While these models excel at general conversational tasks, their static training knowledge limits their ability to recognize domain-specific terms such as contact names, playlists, or technical jargon. Existing solutions primarily rely on prompting, which suffers from poor scalability: as the entity list grows, prompting encounters context window limitations, increased inference latency, and the "lost-in-the-middle" phenomenon. An alternative approach, Generative Error Correction (GEC), attempts to rewrite transcripts via post-processing but frequently suffers from "over-correction", introducing hallucinations of entities that were never spoken.
In this work, we introduce LOGIC (Logit-Space Integration for Contextual Biasing), an efficient and robust framework that operates directly in the decoding layer. Unlike prompting, LOGIC decouples context injection from input processing, ensuring constant-time complexity relative to prompt length. Extensive experiments using the Phi-4-MM model across 11 multilingual locales demonstrate that LOGIC achieves an average 9% relative reduction in Entity WER with a negligible 0.30% increase in False Alarm Rate.
Subjects:
Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Sound (cs.SD)
Cite as:
arXiv:2601.15397 [cs.AI]
(or
arXiv:2601.15397v3 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2601.15397
Focus to learn more
arXiv-issued DOI via DataCite
Submission history
From: Peidong Wang [view email]
[v1]
Wed, 21 Jan 2026 19:08:45 UTC (51 KB)
[v2]
Wed, 4 Feb 2026 20:39:09 UTC (1 KB) (withdrawn)
[v3]
Thu, 10 Sep 2026 17:19:58 UTC (45 KB)
Full-text links:
Access Paper:
View a PDF of the paper titled Beyond Prompting: Efficient and Robust Contextual Biasing for Speech LLMs via Logit-Space Integration (LOGIC), by Peidong Wang and 2 other authors
View PDF
HTML (experimental)
TeX Source
view license
Current browse context:
cs.AI
new
|
recent
| 2026-01
Change to browse by:
cs
cs.CL
cs.SD
References & Citations
NASA ADS
Google Scholar
Semantic Scholar
export BibTeX citation
Loading...
BibTeX formatted citation
×
loading...
Data provided by:
Bookmark
Bibliographic Tools
Bibliographic and Citation Tools
Bibliographic Explorer Toggle
Bibliographic Explorer (What is the Explorer?)
Connected Papers Toggle
Connected Papers (What is Connected Papers?)
Litmaps Toggle
Litmaps (What is Litmaps?)
scite.ai Toggle
scite Smart Citations (What are Smart Citations?)
Code, Data, Media
Code, Data and Media Associated with this Article
alphaXiv Toggle
alphaXiv (What is alphaXiv?)
Links to Code Toggle
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub Toggle
DagsHub (What is DagsHub?)
GotitPub Toggle
Gotit.pub (What is GotitPub?)
Huggingface Toggle
Hugging Face (What is Huggingface?)
ScienceCast Toggle
ScienceCast (What is ScienceCast?)
Demos
Demos
Replicate Toggle
Replicate (What is Replicate?)
Spaces Toggle
Hugging Face Spaces (What is Spaces?)
Spaces Toggle
TXYZ.AI (What is TXYZ.AI?)
Related Papers
Recommenders and Search Tools
Link to Influence Flower
Influence Flower (What are Influence Flowers?)
Core recommender toggle
CORE Recommender (What is CORE?)
Author
Venue
Institution
Topic
About arXivLabs
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Which authors of this paper are endorsers? |
Disable MathJax (What is MathJax?)