analiza aiqbench
This research identifies a specific reasoning bottleneck in Speech Large Language Models where continuous audio features obscure precise entity tracking during logical tasks. The authors demonstrate that an explicit intervention forcing the model to enumerate and bind entities before generating responses can recover significant accuracy, effectively reframing the issue as one of elicitation rather than fundamental capability. This work provides critical diagnostic insight into how speech modality impacts complex reasoning architectures compared to text-only models.
Entity Binding Failures in Speech LLM Reasoning: Diagnosis and Chain-of-Thought Intervention
arXiv:2606.04474v2 Announce Type: replace Abstract: Speech Large Language Models (SLLMs) underperform their text counterparts on complex reasoning. We reveal that this gap is not a uniform cognitive deficit. Evaluating two architecturally diverse SLLMs, we show speech-to-text (S2T) matches or exceeds text-to-text (T2T) on spatial, syntactic, and factual tasks. Yet on logical tasks requiring entity tracking, S2T accuracy collapses to chance. We diagnose this as an entity binding failure: continuous speech features blur precise entity-property associations during implicit reasoning. To validate this diagnosis, we introduce Entity-Aware Chain-of-Thought (EA-CoT), a lightweight inference-time intervention forcing SLLMs to enumerate entities and bind them to claims before reasoning. EA-CoT bridges the gap, even when spoken names are misrecognized, yielding up to a 24.4 percentage-point accuracy gain. Ablations confirm the gains stem from explicit semantic binding, reframing the gap as an elicitation failure rather than a missing capability.
Abstract:Speech Large Language Models (SLLMs) underperform their text counterparts on complex reasoning. We reveal that this gap is not a uniform cognitive deficit. Evaluating two architecturally diverse SLLMs, we show speech-to-text (S2T) matches or exceeds text-to-text (T2T) on spatial, syntactic, and factual tasks. Yet on logical tasks requiring entity tracking, S2T accuracy collapses to chance. We diagnose this as an entity binding failure: continuous speech features blur precise entity-property associations during implicit reasoning. To validate this diagnosis, we introduce Entity-Aware Chain-of-Thought (EA-CoT), a lightweight inference-time intervention forcing SLLMs to enumerate entities and bind them to claims before reasoning. EA-CoT bridges the gap, even when spoken names are misrecognized, yielding up to a 24.4 percentage-point accuracy gain. Ablations confirm the gains stem from explicit semantic binding, reframing the gap as an elicitation failure rather than a missing capability.
Computation and Language (cs.CL); Audio and Speech Processing (eess.AS)
arXiv:2606.04474 [cs.CL]
(or arXiv:2606.04474v2 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2606.04474 Focus to learn more arXiv-issued DOI via DataCite
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.
Text integral sub licența deschisă a publicației · 235 cuvinte
Subiecte conexe
Entity Binding Failures in Speech LLM Reasoning: Diagnosis and Chain-of-Thought Intervention
arXiv · 4 iun. 2026
Underwriting Superintelligence: Backing Agents you can Sue — Rune Kvist, AIUC
Latent Space · 16 sept. 2026
Why I Left Google DeepMind
LessWrong · 15 iul. 2026
I Did This Without Your Knowledge
LessWrong · 17 oct. 2026