AI Navigate

コントラストデコーディングは大規模音声言語モデルをどのように強化するか?

arXiv cs.CL / 2026/3/11

Ideas & Deep AnalysisModels & Research

要点

  • Contrastive Decoding (CD) は大規模音声言語モデル(LALMs)の性能向上に効果があることが示されているが、その成功のメカニズムやさまざまなCD戦略の有効性は以前は明らかでなかった。
  • 本研究では異なるLALMアーキテクチャに対して4つの異なるCD戦略を評価し、Audio-Aware DecodingとAudio Contrastive Decodingが最も効果的な方法であることを特定したが、その効果はモデルによって大きく異なる。
  • 研究者らは推論中の誤りパターンの変化を解析するためにTransition Matrixフレームワークを導入し、CDが音声の不存在を誤って主張する誤りや不確実性に基づく推測を効果的に修正することを明らかにした。
  • しかし、CDは誤った推論や自信を持った誤主張による誤りは修正できず、その補正能力の限界を示している。
  • これらの知見は、基礎となる誤りプロファイルに基づき、どのLALMアーキテクチャがCDによる強化に最適かを判断する実用的な指針を提供する。

Computer Science > Sound

arXiv:2603.09232 (cs)
[Submitted on 10 Mar 2026]

Title:How Contrastive Decoding Enhances Large Audio Language Models?

View a PDF of the paper titled How Contrastive Decoding Enhances Large Audio Language Models?, by Tzu-Quan Lin and 3 other authors
View PDF HTML (experimental)
Abstract:While Contrastive Decoding (CD) has proven effective at enhancing Large Audio Language Models (LALMs), the underlying mechanisms driving its success and the comparative efficacy of different strategies remain unclear. This study systematically evaluates four distinct CD strategies across diverse LALM architectures. We identify Audio-Aware Decoding and Audio Contrastive Decoding as the most effective methods. However, their impact varies significantly by model. To explain this variability, we introduce a Transition Matrix framework to map error pattern shifts during inference. Our analysis demonstrates that CD reliably rectifies errors in which models falsely claim an absence of audio or resort to uncertainty-driven guessing. Conversely, it fails to correct flawed reasoning or confident misassertions. Ultimately, these findings provide a clear guideline for determining which LALM architectures are most suitable for CD enhancement based on their baseline error profiles.
Comments:
Subjects: Sound (cs.SD); Computation and Language (cs.CL); Audio and Speech Processing (eess.AS)
Cite as: arXiv:2603.09232 [cs.SD]
  (or arXiv:2603.09232v1 [cs.SD] for this version)
  https://doi.org/10.48550/arXiv.2603.09232
Focus to learn more
arXiv-issued DOI via DataCite

Submission history

From: Tzu-Quan Lin [view email]
[v1] Tue, 10 Mar 2026 06:05:51 UTC (719 KB)
Full-text links:

Access Paper:

Current browse context:
cs.SD
< prev   |   next >
Change to browse by:

References & Citations

export BibTeX citation Loading...

BibTeX formatted citation

×
Data provided by:

Bookmark

BibSonomy logo Reddit logo
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?)
Links to Code Toggle
Papers with Code (What is Papers with Code?)
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?)
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.