Audio-Conditioned Diffusion LLMs for ASR and Deliberation Processing
Mengqi Wang; Zhan Liu; Zengrui Jin; Guangzhi Sun; Chao Zhang; Philip C. Woodland · 2025 · arXiv
WASTE classifies this as Negative / Null Result Report · AI classification, approximate
The study found no significant effect — useful as a negative control or null benchmark for your own design.
Abstract (excerpt)
Diffusion-based large language models (DLLMs) have recently attracted growing interest as an alternative to autoregressive decoders. In this work, we present an empirical study on using the diffusion-based large language model LLaDA for automatic speech recognition (ASR). We first investigate its use as an external deliberation-based processing module for Whisper-LLaMA transcripts. By leveraging the bidirectional attention and denoising capabilities of LLaDA, we explore random masking, low-confidence masking, and semi-autoregressive strategies, showing that Whisper-LLaDA substantially reduces
Excerpt shown for reference under fair use — read the full paper at the publisher.
About to run something similar?
Run an AI Precheck on your own design to catch failure modes like this one before you spend the time. Your first desk check is free.
Related failures
Aggregation Kinetics of Graphene Oxides in Aqueous Solutions: Experiments, Mechanisms, and Modeling
Negative / Null Result ReportThe reaction between metakaolin and limestone and its effect in porosity refinement and mechanical properties
Negative / Null Result ReportTuning Alginate Bioink Stiffness and Composition for Controlled Growth Factor Delivery and to Spatially Direct MSC Fate within Bioprinted Tissues
Negative / Null Result ReportEx-situ characterisation of gas diffusion layers for proton exchange membrane fuel cells
Negative / Null Result ReportCobb Angle Measurement of Spine from X-Ray Images Using Convolutional Neural Network
Negative / Null Result ReportDeterminants of residential water consumption: Evidence and analysis from a 10‐country household survey
WASTE indexes this work — it does not host or republish it. Failure-type classification is automated and approximate.
Metadata source: arXiv