aiqbench analysis
This research introduces VeriSpeak, a benchmark addressing the critical gap between text-based and speech-based fact verification in Large Audio Language Models. The study demonstrates that while retrieval alone offers limited improvements for spoken claims, combining it with explicit reasoning significantly enhances accuracy to 86.1%. This work underscores the necessity of grounded reasoning over retrieved evidence to effectively combat misinformation in audio formats like podcasts and interviews.
To Trust or Not to Trust: Retrieval-Augmented Fact Checking in Speech
arXiv:2609.30227v1 Announce Type: new Abstract: Online misinformation increasingly appears in spoken formats such as news clips, podcasts, interviews, political speeches, and social media videos, creating a need for fact-checking systems that can verify claims directly from speech. We introduce VeriSpeak, a probe benchmark for studying speech-based fact verification in Large Audio Language Models (LALMs). VeriSpeak contains 3,879 spoken claims spanning temporal, geographical, and relational facts, with balanced true and false labels. The benchmark is designed to examine whether factual verification ability transfers from text to speech, and whether retrieval-augmented LALMs can use textual evidence to correctly support or refute spoken claims. Our experiments reveal a consistent text-speech modality gap: LALMs that verify written claims reliably often fail on the same claims when spoken. Moreover, retrieval alone provides limited gains because models frequently conflate retrieved evidence with the spoken claim. In contrast, retrieval combined with explicit reasoning improves claim-evidence comparison, with a thinking-tuned LALM reaching 86.1% accuracy. VeriSpeak highlights that effective speech misinformation detection requires not only speech understanding, but also grounded reasoning over retrieved evidence. The dataset is publicly available via Hugging Face at https://huggingface.co/datasets/abhiram4572/VeriSpeak.
Abstract:Online misinformation increasingly appears in spoken formats such as news clips, podcasts, interviews, political speeches, and social media videos, creating a need for fact-checking systems that can verify claims directly from speech. We introduce VeriSpeak, a probe benchmark for studying speech-based fact verification in Large Audio Language Models (LALMs). VeriSpeak contains 3,879 spoken claims spanning temporal, geographical, and relational facts, with balanced true and false labels. The benchmark is designed to examine whether factual verification ability transfers from text to speech, and whether retrieval-augmented LALMs can use textual evidence to correctly support or refute spoken claims. Our experiments reveal a consistent text-speech modality gap: LALMs that verify written claims reliably often fail on the same claims when spoken. Moreover, retrieval alone provides limited gains because models frequently conflate retrieved evidence with the spoken claim. In contrast, retrieval combined with explicit reasoning improves claim-evidence comparison, with a thinking-tuned LALM reaching 86.1% accuracy. VeriSpeak highlights that effective speech misinformation detection requires not only speech understanding, but also grounded reasoning over retrieved evidence. The dataset is publicly available via Hugging Face at this https URL.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Sound (cs.SD)
arXiv:2609.30227 [cs.LG]
(or arXiv:2609.30227v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.30227 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
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.
Full text under the publisher's open licence · 285 words
Related coverage
Scaling medical content review at Flo Health with Amazon Bedrock – Part 2
AWS · Jul 14, 2026
In RAG We Trust? Measuring Robustness of Retrieval-Augmented Generation Under Document Poisoning
arXiv · Sep 11, 2026
Building trust in AI agents: A production-ready evaluation framework with Amazon Bedrock AgentCore
AWS · Oct 16, 2026
Natera’s intelligent appointment scheduling with Amazon Bedrock AgentCore
AWS · Aug 26, 2026