arXiv:2608. 09158v1 Announce Type: cross Abstract: Large audio-language models (LALMs) have demonstrated strong capabilities in understanding diverse audio inputs.
arXiv:2607. 03150v1 Announce Type: cross Abstract: While deepfake audio detection systems achieve high performance in controlled benchmarks, their reliability often diminishes in the wild.
arXiv:2609.23416v1 Announce Type: cross Abstract: Long-form audio performance is often summarized by context length and aggregate accuracy, obscuring how language, evidence, and task jointly shape di...
AnchorPrompt is an adaptation technique for large audio‑language models that keeps the base model frozen and learns a single block of prompt vectors inserted at the decoder input. By training these prompts through self‑distillation on diverse audio and text perturbations, the method improves answer consistency and reduces hallucinations across multiple benchmarks. The approach is perturbation‑agnostic at inference, enabling zero‑shot transfer to unseen distortions such as reverberation and choice permutations.
The paper investigates whether audio large language models (Audio LLMs) can detect when their own transcriptions are unreliable. It finds that the models are poor at self-assessment and that existing methods offer limited detection. By leveraging audio-encoder representations, the authors develop a lightweight predictor that accurately flags unreliable transcriptions and can prompt user clarification without altering the underlying model.
arXiv:2606. 17417v1 Announce Type: cross Abstract: Large Audio Language Models (LALMs) achieve strong performance on a variety of audio understanding tasks but continue to struggle with temporal reasoning, a fundamental capability central to human auditory perception.
arXiv:2603. 28378v2 Announce Type: replace-cross Abstract: We present the first systematic Membership Inference Attack (MIA) evaluation of LALMs.
arXiv:2608.22236v2 Announce Type: replace-cross Abstract: Large audio-language models (LALMs) have shown promising progress in understanding speech, music, and general sound events, yet their ability...
The paper introduces a cause-aware error recovery framework for cascaded Automatic Speech Recognition – Large Language Model (ASR‑LLM) pipelines in Spoken Dialogue Systems. It replaces simple ASR confidence filtering with precision‑focused detectors that use deep ASR latent representations to classify token‑level errors into perception, comprehension, and deletion failures. This fine‑grained diagnosis enables the LLM to execute targeted, multi‑turn clarification strategies, leading to a more than two‑fold increase in recall on domain‑shift errors and significant reductions in word error rate and downstream task errors across varied accents, distortions, and domains.
arXiv:2606. 06907v1 Announce Type: cross Abstract: Large audio language models (LALMs) extend large language models with an audio encoder and large-scale audio data.
The paper introduces STAG, a post‑hoc framework that provides token‑level spectro‑temporal grounding for captions produced by audio‑based multimodal large language models (MLLMs). STAG estimates temporal support for each token via vocabulary projections of encoded audio, measures frequency‑band relevance through controlled spectral occlusion, and fuses these signals into a spectro‑temporal relevance map. Evaluations across ten explanation methods and four grounding benchmarks show that STAG achieves superior event‑localization performance on every dataset, and counterfactual deletion experiments confirm that removing the identified evidence selectively reduces model confidence and often eliminates the corresponding event from regenerated captions.
The paper examines why speech‑based screening for Alzheimer’s disease fails to generalize across different languages, tasks, and recording protocols. Using a leave‑one‑corpus‑out evaluation on four datasets, it finds that 59 of 70 interpretable speech features show conflicting patterns between healthy controls and cognitive risk groups, with pause, silence, and speech rate being highly protocol‑sensitive. The authors propose a fusion method that combines XLM‑R text baseline scores with evidence anchors, improving mean speaker AUC to 0.785 and worst‑case AUC to 0.615, and emphasize the importance of auditing feature transferability and reporting worst‑case domain robustness.
arXiv:2608.27783v3 Announce Type: replace-cross Abstract: Speech language models (speech LLMs) can generate plausible outputs from audio that contains no usable speech evidence. We study this failure...