Speaker diarization (SD) is the task of answering "who spoke when" in a multi-speaker audio stream. Classically, an SD system clusters segments of speech belonging to an individual speaker's identity. Recent years have seen substantial progress in SD through end-to-end neural diarization (EEND) approaches. DiariZen, a hybrid SD pipeline built upon a structurally pruned WavLM-Large encoder, a Conformer backend with powerset classification, and VBx clustering, represents the leading open-source state of the art at the time of writing across multiple benchmarks. Despite its strong performance, the DiariZen architecture spans several repositories and frameworks, making it difficult for researchers and practitioners to understand, reproduce, or extend the system as a whole. This tutorial paper provides a self-contained, block-by-block explanation of the complete DiariZen pipeline, decomposing it into seven stages: (1) audio loading and sliding window segmentation, (2) WavLM feature extraction with learned layer weighting, (3) Conformer backend and powerset classification, (4) segmentation aggregation via overlap-add, (5) speaker embedding extraction with overlap exclusion, (6) VBx clustering with PLDA scoring, and (7) reconstruction and RTTM output. For each block, we provide the conceptual motivation, source code references, intermediate tensor shapes, and annotated visualizations of the actual outputs on a 30s excerpt from the AMI Meeting Corpus. The implementation is available at https://github.com/nikhilraghav29/diarizen-tutorial, which includes standalone executable scripts for each block and a Jupyter notebook that runs the complete pipeline end-to-end.

As pretrained large language models replace task-specific decoders in speech recognition, a critical question arises: do their text-derived priors make recognition fairer or more biased across demographic groups? We evaluate nine models spanning three architectural generations (CTC with no language model, encoder-decoder with an implicit LM, and LLM-based with an explicit pretrained decoder) on about 43,000 utterances across five demographic axes (ethnicity, accent, gender, age, first language) using Common Voice 24 and Meta's Fair-Speech, a controlled-prompt dataset that eliminates vocabulary confounds. On clean audio, three findings challenge assumptions: LLM decoders do not amplify racial bias (Granite-8B has the best ethnicity fairness, max/min WER = 2.28); Whisper exhibits pathological hallucination on Indian-accented speech with a non-monotonic insertion-rate spike to 9.62% at large-v3; and audio compression predicts accent fairness more than LLM scale. We then stress-test these findings under 12 acoustic degradation conditions (noise, reverberation, silence injection, chunk masking) across both datasets, totaling 216 inference runs. Severe degradation paradoxically compresses fairness gaps as all groups converge to high WER, but silence injection amplifies Whisper's accent bias up to 4.64x by triggering demographic-selective hallucination. Under masking, Whisper enters catastrophic repetition loops (86% of 51,797 insertions) while explicit-LLM decoders produce 38x fewer insertions with near-zero repetition; high-compression audio encoding (Q-former) reintroduces repetition pathology even in LLM decoders. These results suggest that audio encoder design, not LLM scaling, is the primary lever for equitable and robust speech recognition.

Rings like gold, thuds like wood! The sound we hear in a scene is shaped not only by the spatial layout of the environment but also by the materials of the objects and surfaces within it. For instance, a room with wooden walls will produce a different acoustic experience from a room with the same spatial layout but concrete walls. Accurately modeling these effects is essential for applications such as virtual reality, robotics, architectural design, and audio engineering. Yet, existing methods for acoustic modeling often entangle spatial and material influences in correlated representations, which limits user control and reduces the realism of the generated acoustics. In this work, we present a novel approach for material-controlled Room Impulse Response (RIR) generation that explicitly disentangles the effects of spatial and material cues in a scene. Our approach models the RIR using two modules: a spatial module that captures the influence of the spatial layout of the scene, and a material module that modulates this spatial RIR according to a user-specified material configuration. This explicitly disentangled design allows users to easily modify the material configuration of a scene and observe its impact on acoustics without altering the spatial structure or scene content. Our model provides significant improvements over prior approaches on both acoustic-based metrics (up to +16% on RTE) and material-based metrics (up to +70%). Furthermore, through a human perceptual study, we demonstrate the improved realism and material sensitivity of our model compared to the strongest baselines.

Real-time multimodal agents transport raw audio and screenshots using networking stacks designed for human receivers, which optimize for perceptual fidelity and smooth playout. Yet agent models act as event-driven processors with no inherent sense of physical time, consuming task-relevant semantics rather than reconstructing signals in real time. This fundamental difference shifts the transport goal from the technical problem of signal fidelity (Shannon-Weaver Level A) to the semantic problem of meaning preservation (Level B). This mismatch imposes significant overhead. In visual pipelines, screenshot upload accounts for over 60% of end-to-end action latency on constrained uplinks, and in voice pipelines, conventional transport carries massive redundancy, sending 43-64x more data than needed to maintain task accuracy. We present Sema, a semantic transport system that combines discrete audio tokenizers with a hybrid screen representation (lossless accessibility-tree or OCR text, plus compact visual tokens) and bursty token delivery that eliminates jitter buffers. In simulations under emulated WAN conditions, Sema reduces uplink bandwidth by 64x for audio and 130-210x for screenshots while preserving task accuracy within 0.7 percentage points of the raw baseline.

Quantum algorithms with a proven theoretical speedup over classical computation are rare. Among the most prominent is the Harrow-Hassidim-Lloyd (HHL) algorithm for solving sparse linear systems. Here, HHL is applied to encode melodic preference: the system matrix encodes Narmour implication-realisation and Krumhansl-Kessler tonal stability, so its solution vector is a music-cognition-weighted note-pair distribution. The key constraint of HHL is that reading its output classically cancels the quantum speedup; the solution must be consumed coherently. This motivates a coherent Fourier harmonic oracle: a unitary that applies chord-transition weights directly to the HHL amplitude vector, so that a single measurement jointly selects both melody notes and a two-chord progression. A two-note/two-chord (2/2) block is used to contain the exponential growth of the joint state space that would otherwise make classical simulation of larger blocks infeasible. For demonstrations of longer passages, blocks are chained classically - each block's collapsed output conditions the next -- as a temporary workaround until fault-tolerant hardware permits larger monolithic circuits. A four-block chain produces 8 notes over 8 chords with grammatically valid transitions at every block boundary. Independent rule-based harmony validation confirms that 97% of generated chord progressions are rated strong or acceptable. The primary motivation is that HHL carries a proven exponential speedup over classical linear solvers; this work demonstrates that a coherent HHL+oracle pipeline - the prerequisite for that speedup to be realised in a musical setting - is mechanically achievable. Audio realisations of representative outputs are made available for listening online.

Video-to-music (V2M) is the fundamental task of creating background music for an input video. Recent V2M models achieve audiovisual alignment by typically relying on visual conditioning alone and provide limited semantic and stylistic controllability to the end user. In this paper, we present Video-Robin, a novel text-conditioned video-to-music generation model that enables fast, high-quality, semantically aligned music generation for video content. To balance musical fidelity and semantic understanding, Video-Robin integrates autoregressive planning with diffusion-based synthesis. Specifically, an autoregressive module models global structure by semantically aligning visual and textual inputs to produce high-level music latents. These latents are subsequently refined into coherent, high-fidelity music using local Diffusion Transformers. By factoring semantically driven planning into diffusion-based synthesis, Video-Robin enables fine-grained creator control without sacrificing audio realism. Our proposed model outperforms baselines that solely accept video input and additional feature conditioned baselines on both in-distribution and out-of-distribution benchmarks with a 2.21x speed in inference compared to SOTA. We will open-source everything upon paper acceptance.

Self-supervised learning (SSL) has transformed speech processing, yet its reliance on massive pre-training datasets remains a bottleneck. While robustness is often attributed to scale and diversity, the role of the data distribution is less understood. We systematically examine how curated subsets of pre-training data influence Automatic Speech Recognition (ASR) performance. Surprisingly, optimizing for acoustic, speaker, or linguistic diversity yields no clear improvements over random sampling. Instead, we find that prioritizing the longest utterances achieves superior ASR results while using only half the original dataset, reducing pre-training time by 24% on a large corpora. These findings suggest that for pre-training speech SSL models, data length is a more critical factor than either data diversity or overall data quantity for performance and efficiency, offering a new perspective for data selection strategies in SSL speech processing.

Understanding complete musical scores entails integrated reasoning over pitch, rhythm, harmony, and large-scale structure, yet the ability of Large Language Models and Vision--Language Models to interpret full musical notation remains insufficiently examined. We introduce Musical Score Understanding Benchmark (MSU-Bench), a human-curated benchmark for score-level musical understanding across textual (ABC notation) and visual (PDF) modalities. MSU-Bench contains 1,800 generative question-answer pairs from works by Bach, Beethoven, Chopin, Debussy, and others, organised into four levels of increasing difficulty, ranging from onset information to texture and form. Evaluations of more than fifteen state-of-the-art models, in both zero-shot and fine-tuned settings, reveal pronounced modality gaps, unstable level-wise performance, and challenges in maintaining multilevel correctness. Fine-tuning substantially improves results across modalities while preserving general knowledge, positioning MSU-Bench as a robust foundation for future research in multimodal reasoning. The benchmark and code are available at https://github.com/Congren-Dai/MSU-Bench.

The rapid development of Artificial Intelligence Generated Content (AIGC) has made high-fidelity generated audio widely available across the Internet, driving the advancement of audio steganography. Benefiting from advances in deep learning, current audio steganography schemes are mainly based on encoder-decoder network architectures. While these methods guarantee a certain level of perceptual quality for stego audio, they typically face high computational cost and long implementation time, as well as poor anti-steganalysis performance. To address the aforementioned issues, we pioneer a Fixed Decoder Network-Based Audio Steganography with Adversarial Perturbation Generation (FGAS). Adversarial perturbations carrying a secret message are embedded into the cover audio to generate stego audio. The receiver only needs to share the structure and key of the fixed decoder network to accurately extract the secret message from the stego audio. In FGAS, we propose an Audio Adversarial Perturbation Generation (A2PG) strategy with an optional robust extension and design a lightweight fixed decoder. The fixed decoder guarantees reliable extraction of the hidden message, while adversarial perturbations are optimized to keep the stego audio perceptually and statistically close to the cover audio, thereby improving anti-steganalysis performance. The experimental results show that FGAS significantly improves stego audio quality, achieving an average PSNR gain of over 10 dB compared to SOTA methods. Furthermore, FGAS demonstrates strong robustness against common audio processing attacks. Moreover, FGAS exhibits superior anti-steganalysis performance across different relative payloads; under high-capacity embedding, it achieves a classification error rate about 2% higher, indicating stronger anti-steganalysis performance than current SOTA methods.

Dementia encompasses a group of syndromes that impair cognitive functions such as memory, reasoning, and the ability to perform daily activities. As populations globally age, over 10 million new dementia diagnoses are reported annually. Currently, clinical diagnosis of dementia remains challenging due to overlapping symptoms, the need to exclude alternative conditions and the requirement for a comprehensive clinical evaluation and cognitive assessment. This underscores the growing need to develop feasible and accurate methods for detecting cognitive deficiencies. Recent advances in machine learning have highlighted spontaneous speech as a promising noninvasive, cost-effective, and scalable biomarker for dementia detection. In this study, spontaneous speech recordings from the ADReSS and Pitt Corpus datasets are analyzed, consisting of picture description tasks performed by cognitively healthy individuals and people with Alzheimer's disease. Unlike prior approaches that focus solely on speech-active segments, acoustic features are extracted from entire recordings using the openSMILE toolkit. This representation reduces the number of feature vectors and improves computational efficiency without compromising classification performance. Classification models with classifier-based wrapper feature selection are employed to estimate feature importance and identify diagnostically relevant acoustic characteristics. Among the evaluated models, the Extreme Minimal Learning Machine achieved competitive classification accuracy with substantially lower computational cost, reflecting an inherent property of the model formulation and learning procedure. Overall, the results demonstrate that the proposed framework is computationally efficient, interpretable, and well suited as a supportive tool for speech-based dementia assessment.

Computational models of syntax are predominantly text-based. Here we propose that the most basic first step in the evolution of syntax can be modeled directly from raw speech in a fully unsupervised way. We focus on one of the most ubiquitous and elementary suboperations of syntax -- concatenation. We introduce \textit{spontaneous concatenation}: a phenomenon where a ciwGAN/fiwGAN models (based on convolutional neural networks) trained on acoustic recordings of individual words start generating outputs with two or even three words concatenated without ever accessing data with multiple words in the training data. We replicate this finding in several independently trained models with different hyperparameters and training data. Additionally, networks trained on two words learn to embed words into novel unobserved word combinations. We also show that the concatenated outputs contain precursors to compositionality. To our knowledge, this is a previously unreported property of CNNs trained in the ciwGAN/fiwGAN setting on raw speech and has implications both for our understanding of how these architectures learn as well as for modeling syntax and its evolution in the brain from raw acoustic inputs. We also propose and formalize a neural mechanism called \textit{disinhibition} that outlines a possible artificial and biological neural pathway towards concatenation and compositionality and suggests our modeling is useful for generating testable predictions for biological and artificial neural processing of spoken language.