MANAR: Memory-augmented Attention with Navigational Abstract Conceptual Representation

MANAR (Memory-augmented Attention with Navigational Abstract Conceptual Representation), contextualization layer generalizes standard multi-head attention (MHA) by instantiating the principles of Global Workspace Theory (GWT). While MHA enables unconstrained all-to-all communication, it lacks the functional bottleneck and global integration mechanisms hypothesized in cognitive models of consciousness. MANAR addresses this by implementing a central workspace through a trainable memory of abstract concepts and an Abstract Conceptual Representation (ACR). The architecture follows a two-stage logic that maps directly to GWT mechanics: (i) an integration phase, where retrieved memory concepts converge to form a collective "mental image" (the ACR) based on input stimuli; and (ii) a broadcasting phase, where this global state navigates and informs the contextualization of individual local tokens. We demonstrate that efficient linear-time scaling is a fundamental architectural byproduct of instantiating GWT functional bottleneck, as routing global information through a constant-sized ACR resolves the quadratic complexity inherent in standard attention. MANAR is a compatible re-parameterization of MHA with identical semantic roles for its projections, enabling knowledge transfer from pretrained transformers via weight-copy and thus overcoming the adoption barriers of structurally incompatible linear-time alternatives. MANAR enables non-convex contextualization, synthesizing representations that provably lie outside the convex hull of input tokens - a mathematical reflection of the creative synthesis described in GWT. Empirical evaluations confirm that MANAR matches or exceeds strong baselines across language (GLUE score of 85.1), vision (83.9% ImageNet-1K), and speech (2.7% WER on LibriSpeech), positioning it as an efficient and expressive alternative to quadratic attention.

On Optimizing Multimodal Jailbreaks for Spoken Language Models

As Spoken Language Models (SLMs) integrate speech and text modalities, they inherit the safety vulnerabilities of their LLM backbone and an expanded attack surface. SLMs have been previously shown to be susceptible to jailbreaking, where adversarial prompts induce harmful responses. Yet existing attacks largely remain unimodal, optimizing either text or audio in isolation. We explore gradient-based multimodal jailbreaks by introducing JAMA (Joint Audio-text Multimodal Attack), a joint multimodal optimization framework combining Greedy Coordinate Gradient (GCG) for text and Projected Gradient Descent (PGD) for audio, to simultaneously perturb both modalities. Evaluations across four state-of-the-art SLMs and four audio types demonstrate that JAMA surpasses unimodal jailbreak rate by 1.5x to 10x. We analyze the operational dynamics of this joint attack and show that a sequential approximation method makes it 4x to 6x faster. Our findings suggest that unimodal safety is insufficient for robust SLMs. The code and data are available at https://repos.lsv.uni-saarland.de/akrishnan/multimodal-jailbreak-slm

Empathetic Motion Generation for Humanoid Educational Robots via Reasoning-Guided Vision--Language--Motion Diffusion Architecture

This article suggests a reasoning-guided vision-language-motion diffusion framework (RG-VLMD) for generating instruction-aware co-speech gestures for humanoid robots in educational scenarios. The system integrates multi-modal affective estimation, pedagogical reasoning, and teaching-act-conditioned motion synthesis to enable adaptive and semantically consistent robot behavior. A gated mixture-of-experts model predicts Valence/Arousal from input text, visual, and acoustic features, which then mapped to discrete teaching-act categories through an affect-driven policy.These signals condition a diffusion-based motion generator using clip-level intent and frame-level instructional schedules via additive latent restriction with auxiliary action-group supervision. Compared to a baseline diffusion model, our proposed method produces more structured and distinctive motion patterns, as verified by motion statics and pairwise distance analysis. Generated motion sequences remain physically plausible and can be retargeted to a NAO robot for real-time execution. The results reveal that reasoning-guided instructional conditioning improves gesture controllability and pedagogical expressiveness in educational human-robot interaction.

"You've got a friend in me": Co-Designing a Peer Social Robot for Young Newcomers' Language and Cultural Learning

Community literacy programs supporting young newcomer children in Canada face limited staffing and scarce one-to-one time, which constrains personalized English and cultural learning support. This paper reports on a co-design study with United for Literacy tutors that informed Maple, a table-top, peer-like Socially Assistive Robot (SAR) designed as a practice partner within tutor-mediated sessions. From shadowing and co-design interviews, we derived newcomer-specific requirements and added them in an integrated prototype that uses short story-based activities, multi-modal scaffolding (speech, facial feedback, gesture), and embedded quizzes that support attention while producing tutor-actionable formative signals. We contribute system design implications for tutor-in-the-loop SARs supporting language socialization in community settings and outline directions for child-centered evaluation in authentic programs.

ARTT: Augmented Reverberant-Target Training for Unsupervised Monaural Speech Dereverberation

Due to the absence of clean reference signals and spatial cues, monaural unsupervised speech dereverberation is a challenging ill-posed inverse problem. To realize it, we propose augmented reverberant-target training (ARTT), which consists of two stages. In the first stage, reverberant-target training (RTT) is proposed to first further reverberate the observed reverberant mixture signal, and then train a deep neural network (DNN) to recover the observed reverberant mixture via discriminative training. Although the target signal to fit is reverberant, we find that the resulting DNN can effectively reduce reverberation. In the second stage, an online self-distillation mechanism based on the mean-teacher algorithm is proposed to further improve dereverberation. Evaluation results demonstrate that ARTT achieves strong unsupervised dereverberation performance, significantly outperforming previous baselines.

Enhancing Multi-Corpus Training in SSL-Based Anti-Spoofing Models: Domain-Invariant Feature Extraction

The performance of speech spoofing detection often varies across different training and evaluation corpora. Leveraging multiple corpora typically enhances robustness and performance in fields like speaker recognition and speech recognition. However, our spoofing detection experiments show that multi-corpus training does not consistently improve performance and may even degrade it. We hypothesize that dataset-specific biases impair generalization, leading to performance instability. To address this, we propose an Invariant Domain Feature Extraction (IDFE) framework, employing multi-task learning and a gradient reversal layer to minimize corpus-specific information in learned embeddings. The IDFE framework reduces the average equal error rate by 20% compared to the baseline, assessed across four varied datasets.

DiscoPhon: Benchmarking the Unsupervised Discovery of Phoneme Inventories With Discrete Speech Units

We introduce DiscoPhon, a multilingual benchmark for evaluating unsupervised phoneme discovery from discrete speech units. DiscoPhon covers 6 dev and 6 test languages, chosen to span a wide range of phonemic contrasts. Given only 10 hours of speech in a previously unseen language, systems must produce discrete units that are mapped to a predefined phoneme inventory, through either a many-to-one or a one-to-one assignment. The resulting sequences are evaluated for unit quality, recognition and segmentation. We provide four pretrained multilingual HuBERT and SpidR baselines, and show that phonemic information is available enough in current models for derived units to correlate well with phonemes, though with variations across languages.

Listen First, Then Answer: Timestamp-Grounded Speech Reasoning

Large audio-language models (LALMs) can generate reasoning chains for their predictions, but it remains unclear whether these reasoning chains remain grounded in the input audio. In this paper, we propose an RL-based strategy that grounds the reasoning outputs of LALMs with explicit timestamp annotations referring to relevant segments of the audio signal. Our analysis shows that timestamp grounding leads the model to attend more strongly to audio tokens during reasoning generation. Experiments on four speech-based benchmark datasets demonstrate that our approach improves performance compared to both zero-shot reasoning and fine-tuning without timestamp grounding. Additionally, grounding amplifies desirable reasoning behaviors, such as region exploration, audiology verification, and consistency, underscoring the importance of grounding mechanisms for faithful multimodal reasoning.

EARTalking: End-to-end GPT-style Autoregressive Talking Head Synthesis with Frame-wise Control

Audio-driven talking head generation aims to create vivid and realistic videos from a static portrait and speech. Existing AR-based methods rely on intermediate facial representations, which limit their expressiveness and realism. Meanwhile, diffusion-based methods generate clip-by-clip, lacking fine-grained control and causing inherent latency due to overall denoising across the window. To address these limitations, we propose EARTalking, a novel end-to-end, GPT-style autoregressive model for interactive audio-driven talking head generation. Our method introduces a novel frame-by-frame, in-context, audio-driven streaming generation paradigm. For inherently supporting variable-length video generation with identity consistency, we propose the Sink Frame Window Attention (SFA) mechanism. Furthermore, to avoid the complex, separate networks that prior works required for diverse control signals, we propose a streaming Frame Condition In-Context (FCIC) scheme. This scheme efficiently injects diverse control signals in a streaming, in-context manner, enabling interactive control at every frame and at arbitrary moments. Experiments demonstrate that EARTalking outperforms existing autoregressive methods and achieves performance comparable to diffusion-based methods. Our work demonstrates the feasibility of in-context streaming autoregressive control, unlocking a scalable direction for flexible, efficient generation. The code will be released for reproducibility.

Zipper-LoRA: Dynamic Parameter Decoupling for Speech-LLM based Multilingual Speech Recognition

Speech Large Language Models (Speech-LLMs) have emerged as a powerful approach for automatic speech recognition (ASR) by aligning speech encoders with large language models. However, adapting these systems to multilingual settings with imbalanced data distributions remains challenging. In such scenarios, a stability-plasticity dilemma often arises: fully shared Parameter-Efficient Fine-Tuning (PEFT) can cause negative inter-lingual interference for under-represented languages, while fully language-specific tuning limits the cross-lingual beneficial knowledge transfer needed for low-resource tasks. To address this, we propose Zipper-LoRA, a novel rank-level decoupling framework with three variants (Static, Hard, and Soft) that dynamically synthesizes LoRA updates from shared and language-specific subspaces. By using a lightweight language-conditioned router, Zipper-LoRA dynamically controls the contribution of each subspace at the LoRA rank level, enabling fine-grained sharing where languages are compatible and strict decoupling when conflicts occur. To further stabilize optimization under imbalanced data, we propose a two-stage training strategy with an Initial-B warm start that significantly accelerates convergence. Experiments on a 12-language mixed-resource setting show that Zipper-LoRA consistently outperforms both fully shared and independent baselines, particularly in extremely low-resource scenarios. Moreover, we demonstrate that these gains are robust across both chunked and non-chunked encoder configurations, confirming the framework's reliability for practical, large-scale multilingual ASR. Our code and data will be available at https://github.com/YuCeong-May/Zipper-LoRA for reproducibility.