Papers

Speech Language Models (SLMs) are increasingly deployed in multi-speaker environments, yet their ability to attribute speech to the correct speaker and reason over speaker identities remains unclear. Hence, we introduce HEAR, a conceptually hierarchical benchmark diagnosing the foundational capabilities of speaker-attributed reasoning, comprising 2.4K human-verified samples from 887 diverse multi-party audio clips. Evaluating 20 leading SLMs on HEAR reveals they struggle with these foundational tasks, often relying on semantic priors rather than actual vocal cues. To address this, we present A2R, a 30B model optimized on Counterfactual Audio with Speaker-level Hard negatives (CASH), a dataset designed to guide the model to prioritize acoustic vocal cues over linguistic signals. A2R achieves strong performance on HEAR and exhibits zero-shot generalization to diverse multi-speaker downstream tasks, demonstrating that learned speaker attribution unlocks the model's latent capacity for speaker-aware reasoning. All resources are available at https://attributetoreason.github.io/AttributeToReason/

Current TTS systems typically rely on open-loop, single-pass generation and can produce sporadic local prosodic defects, such as misplaced stress, unnatural pauses, or flattened intonation, that utterance-level metrics often fail to expose. We present LoopTTS, a judge-guided Filter-Judge-Refiner framework for recovering low-quality TTS outputs diagnosed by an AudioLLM. Given an initial utterance from a base TTS model, an AudioLLM Judge identifies salient prosodic issues and generates structured refine instructions; a Refiner, our fine-grained instruction-following TTS model, then performs guided expressive re-synthesis conditioned on the initial utterance, target text, and instruction. To train the Refiner, we construct Refiner-DB, a 42K-example AudioLLM-annotated dataset with word-level prosodic weak supervision. Human evaluation on diagnosed low-quality utterances shows that LoopTTS can detect perceptually salient errors and correct them with the Refiner, outperforming raw generated audio and practical open-loop re-generation baselines in recovery quality. The Refiner also demonstrates stronger instruction-following ability for stress and pause control in targeted prosody modification.

An avatar that holds a conversation should decide what to say and to move while saying it, yet these abilities live in separate model families: spoken dialogue models produce speech without motion, and co-speech motion models produce motion only from audio handed to them. The standard remedy is a cascade that first generates the spoken response and then runs a motion model over the finished audio, which requires a second full inference pass and precludes any joint optimisation between the two. We present Motion-Omni, an end-to-end framework in which a spoken dialogue model natively outputs explicit facial expression together with hand, upper-body and lower-body motion, generated directly from the hidden states that produce the speech. Joint training is not optional here: with the speech pathway frozen, motion remains misaligned with the audio, and co-adapting the LLM, Speech Generator and Motion Generator under both objectives is what recovers alignment while retaining spoken-dialogue ability. Supervision comes from a scalable, model-agnostic pipeline that pseudo-labels consistent-voice speech responses with a replaceable motion teacher, yielding 422,856 quality-ranked pairs (1,402 hours). We further release SwDA-500 and, to our knowledge, the first public evaluation protocol for stochastic open-ended full-body spoken dialogue, matching audio across motion systems while unifying rendering, automatic metrics, human evaluation, and latency measurement. Instantiated with a Qwen2.5-7B-Instruct backbone, Motion-Omni-Q7 matches the same-audio teacher cascade to within 2% on reference-free motion metrics while responding 5.4 x faster (RTF=0.78, faster than real time), surpasses all non-teacher cascades on beat correlation and diversity, and reaches a 2.62% word error rate, the lowest among the omni-modal systems compared.

Existing audio-video generation methods produce well-synchronized clips but are limited to short durations, while long-video generation methods extend duration through chunk-based iterative synthesis yet lack audio entirely. Generating long audio-video jointly is fundamentally harder than either task alone: each chunk must simultaneously maintain video temporal coherence, audio temporal coherence, and cross-modal synchronization, whose conditioning signals enter the model through different pathways. In this work, we present Encore for long-form synchronized audio-video generation. Our key insight is to factor this challenge into: (1) local continuity handled by iterative generation with explicit cross-chunk context propagation, and (2) global consistency enforced via reference audio-video signals with shifted position embedding. Building on this design, we propose Adaptive Signal Routing (ASR), which introduces learnable attention biases within self-attention and learnable residual scales on cross-attention outputs, enabling the model to adaptively modulate the influence of each conditioning signal. Trained end-to-end for joint audio-video generation, Encore also supports infinite-length audio-to-video and video-to-audio synthesis at inference by conditioning on the ground-truth modality throughout the denoising process. Experiments on our extended VerseBench for long audio-video evaluation demonstrate that Encore significantly outperforms existing methods in both generation quality and temporal coherence. Code and data for this paper are at https://github.com/shaohua-pan/Encore.

Voice products increasingly need affective cues that are present in speech but absent from transcripts. We introduce VocalAffectBench, a public, test-only benchmark for evaluating whether AI audio models can identify expressed vocal emotion from raw audio. The benchmark contains 273 human-recorded English WAV clips from 51 speaker accounts totaling 1.95 hours across seven labels: angry, disgusted, fearful, happy, neutral, sad, and surprised, with 39 clips per class. All baselines are evaluated from audio alone, without transcripts or contextual metadata. Across six released baselines, average accuracy is 35.5%. The strongest baseline, gemini_3_5_flash, reaches 46.5% on the seven-way task, above the 14.3% random baseline but far from robust emotion recognition. A secondary valence-bucket analysis maps labels into positive, neutral, and negative classes, excluding surprised because its valence is ambiguous. Aggregate accuracy under this coarser view is 50.9%. Performance is highly uneven across classes. By recall, neutral is identified most reliably at 75.6% averaged across baselines, while surprised and fearful reach only 10.7% and 15.4%, respectively. These results show that the evaluated baselines can extract some affective signal from speech, but discrete expressed-emotion recognition remains fragile, especially for non-neutral emotions that are often most important in voice agent workflows.

Supplying context at inference time to a large multimodal model is an inexpensive lever for adapting speech transcription to a domain, and earlier results on smaller models reported large gains. This work tested that mechanism where it ships, in the prompt-conditioning layer of a production oral-history transcription tool, on a sample from its own production corpus. Full prompt-level context did not detectably change side-level word error rate (WER), and none of the four preregistered hypotheses was supported. The design was a within-item paired ablation, preregistered with the analysis code frozen by hash before the confirmatory batch was scored; two disclosed gpt-4o pilot sides had been scored earlier, during scorer development. Nineteen cassette sides, about 10.6 hours of degraded 1970s-80s interview audio, were reprocessed through the production code path under three prompt arms, crossed with two deployed commercial configurations, gpt-4o-transcribe and gemini-2.5-flash, and scored against operator-corrected verbatim references. For gpt-4o-transcribe the median paired difference between the full-context and no-context arms was +0.6 WER points, with a side-resampled interval of [-1.1, +1.0]; the Gemini estimates were too unstable to support a comparable negative inference. A post-hoc rerun found run-to-run pipeline variability larger than the confirmatory differences, so effects of that size cannot be resolved from one transcription per cell. An implementation audit verified the manipulation was live, and sequence-alignment analysis found a small improvement on complete context-listed phrases, too small to materially change side-level WER, and for Gemini coexisting with worsened unlisted-token error. Evaluating context mechanisms therefore requires sequence-aligned term-level, insertion, and speaker-label measures alongside aggregate accuracy.

I show how to evoke the pitch-class content of a chord from an arbitrary source sound by convolving the source with an impulse response whose grains are one windowed sinusoid per pitch-class, across each octave of hearing range; while, at the same time, minimizing artifacts. A Csound user-defined opcode, chord_convolver, mixes a dry Dirac component into that response, and applies partitioned convolution once. I contrast the effect with a linear-frequency comb filter and with a generic constant-Q resonator bank, and I demonstrate musical use on a twilight field recording alongside the ruins of Chateau de Lagarde.

We describe our submission to Task A of the 1st DAFx parameter estimation challenge. The task is to recover the six physical parameters of a simulated metal-plate reverberator – its dimensions and material properties – from a single impulse response (IR). We treat this as a black-box optimization: candidate parameter sets are fed to the simulator and scored by a loss against the target IR. The method has two stages. The first uses CMA-ES, an evolutionary optimizer, to recover five of the six parameters, comparing IRs under an amplitude-normalized loss. Amplitude normalization makes the search robust but discards the cue to the sixth parameter, the plate's surface density; a second stage therefore estimates it alone, with a ternary search on the un-normalized loss. As the choice of loss strongly affects the search, we select it beforehand, and analyze why compression in the common multi-scale spectral loss degrades recovery. Finally, we test our method on a validation set of 50 IRs, discuss a pathological failure mode, and ablate to justify having two different stages instead of a unified CMA-ES search.

Cochlear implants (CI) restore hearing for individuals with severe to profound hearing loss. However, CI users often struggle to understand speech in noisy environments. Deep neural networks (DNN) have shown promise in enhancing speech for CI users, yet their high energy demands make them non-ideal for low-power CI processors. Spiking neural networks (SNN), on the other hand, offer comparable performance with significantly lower energy consumption. Hence, we propose a novel SNN inspired by the Deep ACE architecture that simultaneously performs speech enhancement and CI coding. Our model achieves competitive vocoded short-time objective intelligibility (VSTOI) and signal-to-noise ratio improvement (SNRi) scores compared to Deep ACE, while achieving more than a sixfold reduction in energy consumption.

Deep neural networks (DNNs) based on state-space models (SSMs) are increasingly applied to speech processing, but typically operate on pulse-code-modulated (PCM) audio. This constrains deployment on low-power, always-on edge devices, which commonly use single-bit pulse-density-modulated (PDM) micro-electromechanical (MEMS) microphones for their noise robustness, low cost, and variable sampling rates that enable low-power operation. In fact, converting PDM to PCM requires low-pass filtering and decimation, imposing costly overhead on resource-constrained hardware. While prior works have attempted to process PDM signals directly, they require long training times and generalize poorly across sampling rates. In this paper, we show that the SSM has two key properties that remediate these issues: its continuous-time parametrization allows it to produce a consistent representation of the input audio signal, regardless of the modulation strategy and sampling rate, and its long-term memory enables this representation to be aggressively downsampled without needing any anti-aliasing operations. We then propose a novel end-to-end PDM speech processing architecture that uses an SSM to encode the input audio signal into a modulation- and sampling-rate-invariant latent representation. We show that our proposed architecture achieves robust speech classification and enhancement gains at low-power sampling-rates (512 kHz) and similar performance to state-of-the-art algorithms operating on PCM data when tested on standard PDM sampling-rates of 2 MHz. Moreover, we show that the SSM's output can be downsampled by more than 65,000 times, thus significantly reducing the number of processing timesteps in downstream layers.

Musicians commonly communicate music through scores and performances. Scores encode musical intent, while performances realize it in sound. To investigate whether models can meaningfully engage with both modalities, we introduce MuSP-Bench, a human-authored benchmark of 490 questions targeting understanding across Musical Scores and Performances. The benchmark distinguishes itself by spanning score-based, performance-based, interpretive, and long-horizon reasoning across classical piano and orchestral works. We evaluate frontier multimodal large language models under multiple input conditions. Our results show that these models struggle substantially to understand scores, while facing even greater challenges when reasoning about performance audio. The benchmark is available at https://musp.vaclis.net/.

Neural audio representation learning has enabled a range of content-oriented applications, but the resulting features remain limited for tasks involving audio processing. Furthermore, it is not obvious what processing-aware representations should capture: the processing itself, abstracted away from source content, or the processed audio that retains it. Existing approaches implicitly commit to one or the other and also differ in their models, data, and evaluation, obscuring which design choices drive their behavior. We address both questions within a unified framework of three objectives: processing consistency, description alignment, and equivariance via forward prediction. We compare all combinations of the objectives under a controlled setup and reveal their relative strengths and interactions. Our framework produces both a transformation embedding and a processed-audio embedding, and we find that the two play complementary roles: distance-based tasks favor the former, while probe-based tasks favor the latter. Combined with improvements in network architecture and training pipeline, our representations outperform prior baselines across retrieval, probe-based evaluation, and style transfer.

Existing approaches to evasion detection in earnings calls focus on textual transcripts, treating evasion as a single-dimensional phenomenon. We argue that evasion in spoken communication is inherently multidimensional: beyond what executives say, how they say it carries independent and complementary information. To study these dimensions jointly, we introduce DualEvasion, a benchmark for evasion detection across text and audio in earnings call Q&A. The benchmark contains 505 annotated question-answer pairs from 60 earnings calls, each with two independent labels: textual evasion (direct vs. evasive) and vocal cues operationalized as speaker confidence (confident vs. unconfident). Our experiments show that state-of-the-art multimodal models struggle to detect vocal confidence, particularly on unconfident responses. Our analysis suggests these models interpret acoustic cues in isolation rather than relative to each speaker's baseline. Providing speaker-level references yields modest improvements, but a substantial gap with human performance remains.

This paper deals with the introduction of "geometric harmony", a discipline that explicitly addresses the spectral characteristics of musical gamut. The framework of Western music, from Renaissance to the present, represents sound in terms of "pitch"-as is evident from its five-line staff notation system-and employs the fundamental frequency as its representative value, 440 Hz, etc. In this paper, by taking the timbres of specific individual instruments as elements and examining the Wasserstein distance between two voices, and Wasserstein deviations between three or more voices, we demonstrate that it is possible to expand the system whilst retaining the entire framework of conventional music theory. At the same time, as an example of practical utility in ensemble playing, we provide a detailed account of the two-voices affinity of the "Throat G" on clarinet, a note known for its fragility in ensemble contexts.

Smooth speaker transitions are fundamental to effective conversation and rely on an interlocutor's ability to predict when to enter the conversation. This ability depends on accurately interpreting and expressing the verbal and non-verbal cues that signal when a speaker wishes to take or relinquish the floor. The process becomes even more complex in noisy, natural, multi-party settings, with multiple interlocutors available. This study models how gaze and speech, together with perceived interpersonal closeness, signal conversational floor changes in free four-person dialogue. Using the GaMMA corpus, we trained logistic regression models using interpretable, behaviourally motivated features extracted before each turn-taking event to classify floor-transfer outcomes as gaps or overlaps. Predictors included gaze features such as transition motifs and behavioural contrasts, entropy, gaze-based addressee identity, and mutual gaze, alongside speech features derived from speaker loudness, as well as perceived interpersonal closeness (IOS) between speakers. Results show that gaze features capture predictive structure, and that combining them with loudness improves performance (ROC AUC = 0.76 +- 0.04). Loudness reflected speaker control, while gaze dispersion and addressing indexed listener readiness and competitive entry. Performance remained robust across noise conditions, indicating that gaze provides a complementary, noise-resilient cue to turn-taking dynamics.

Prosody plays an important role in speech translation, conveying information such as emphasis, emotion, and intent beyond lexical content. However, despite recent progress in expressive speech-to-speech translation (S2ST), little is known about how prosodic patterns are similar/different across languages. Understanding these cross-lingual similarities and differences is crucial for effectively incorporating prosody into expressive S2ST systems. In this work, we present the first fine-grained cross-lingual analysis of prosody using multilingual dubbing data across English-German, English-Spanish, and English-French language pairs. We analyze the similarity of pitch, energy, and temporal feature patterns between source and target speech and investigate the linguistic and alignment-related factors affecting this similarity. Our analysis reveals inherent cross-lingual correlations in prosodic structure between certain languages. The findings provide important insights into the transferability of prosody across languages and offer empirical guidance for future expressive speech-to-speech translation systems.

Speech and audio LLMs are evaluated by comparing waveform predictions with predictions from an automatic speech recognition (ASR) transcript. For fixed closed-set tasks, this conflates acoustic evidence with the need to invoke a generative audio model. We estimate incremental call value with matched selectors sharing pre-call evidence. Each policy may retain the transcript label, use a local encoder, or invoke a generative model; matched control removes generative actions but preserves pre-call evidence and development selection. On VocalSound, transcript-only accuracy is 0.296, while supervised CLAP and WavLM controls reach 0.850 and 0.854 without calls. Full selector reaches 0.925 at 12.5% calls versus 0.921 for matched No-call selector (difference 0.004; 95% CI [-0.025, 0.033]). Thus, results do not show a call gain after transcript and encoder evidence are available. Relevant quantity is incremental accuracy from allowing calls, not the waveform-transcript gap.

Speech language models (speech LLMs) can generate plausible outputs from audio that contains no usable speech evidence. We study this failure as a pre-generation support-estimation problem and present SURE-Voice, a training-free front end that decides whether an audio prompt contains intelligible speech evidence before calling a speech LLM. We build SURE-Challenge with a 640-example SURE-Core split and a 1,920-example SURE-Extended split derived from 120 LibriSpeech source utterances. Using one fixed operating point, an energy screen plus Whisper token confidence raises unsupported accuracy on the held-out Extended test from 0.000–0.133 to 0.919 for six non-degenerate speech LLM backbones, while supported accuracy remains 0.919–0.970 and downstream calls fall from 480 to 287. A 500-clip ESC-50 sanity set shows the same pattern on real environmental audio, with vocal non-speech as a residual failure mode. An overlap diagnostic shows that source attribution remains separate from speech-evidence filtering. The evidence supports a controlled benchmark baseline and a deployment-oriented analysis; it does not establish universal robustness to semantic answerability, gain variation or natural conversations.

The relationship between subjective test vote variance and vote mean (or MOS) is well-studied, and the mathematically admissible vote variance region has been previously defined. We propose a reduced admissible variance region called the Unimodal Variance Region (UVR) that better describes real subjective rating behavior of multimedia. Further, subjective vote variance is often modeled as parabolic. We explain that, in practice, the parabolic model often violates the admissible region in the variance vs. MOS plane and we propose alternatives that respect the admissible region. We also present a parametrized random process to model votes that mixes voting processes and produces a realistic range of vote variances within the UVR at any desired MOS. This process was inspired by and comports with voting behavior that is observed in many subjective tests. By modeling vote variance from a subjective experiment, this vote model offers additional interpretable insights into voting behavior observed in a given experiment. We present example results from 16 datasets spanning speech, image, and video subjective quality experiments.

In out-of-domain (OOD) sound-matching, a synthesizer is optimized to mimic a sound it did not generate. OOD evaluation of loss functions is underexplored in part because the standard "parameter loss" metric requires a shared parameter space between target and imitator, which OOD settings lack. We introduce Partial Parameter Distance (PPD), which applies parameter loss only to the critical parameters that mismatched synthesizers share (e.g., filter cutoffs), enabling automatically evaluated OOD experiments; we verify its results with blinded listening tests. Across seven scenarios involving band-pass filtering, amplitude modulation, and pitch-bending, we evaluate four differentiable loss functions (SIMSE_Spec, L1_Spec, JTFS, DTW_Envelope). Loss-function effectiveness remains tightly coupled to the method of synthesis: SIMSE_Spec excels at filter-cutoff recovery, DTW_Envelope at amplitude-modulation recovery, and JTFS at smooth pitch trajectories. Parameter-based evaluation agrees with listening tests on the top-ranked loss function in five of seven scenarios, demonstrating its utility as a diagnostic tool.

Silent speech recognition (SSR) provides an alternative communication pathway in the absence of audible speech. However, conventional approaches are limited by the need for constant facial attachment, privacy concerns, and unstable signal acquisition. Here, we propose a soft, active electromyography (EMG) interface that enables word-level SSR using machine learning. Worn on the hand, the device uses a fingertip electrode that can be positioned near the lips to acquire EMG signals only when needed. The interface integrates liquid metal (LM) interconnects, transparent flexible printed circuit (FPC) electrodes, and elastomer encapsulation to ensure high mechanical stability during finger motion. A deep neural network trained on these stable signals achieved a mean accuracy of 97.2 $\pm$ 1.3% across three subjects in classifying a 30-word vocabulary, demonstrating robust linguistic discrimination. Furthermore, real-time drone control validates the practicality of this approach in noisy and privacy-sensitive environments where conventional voice recognition fails. This study highlights the potential of soft, wearable EMG systems as secure and intuitive human-machine interfaces.

Large Audio Language Models (LALMs) have demonstrated strong performance across a wide range of audio understanding tasks. However, they are typically evaluated on single, coherent audio segments, leaving their behavior under less familiar input configurations underexplored. We study this issue through a controlled setting in which two audio segments are concatenated into a single input. Across multiple LALMs, we observe a striking task-dependent robustness gap: automatic speech recognition (ASR) remains comparatively stable, whereas audio question answering (AQA) degrades substantially. To investigate the mechanisms underlying this disparity, we analyze how audio information is routed through LALM decoders using layer-wise attention knockout. The results reveal distinct task-dependent pathways. ASR relies primarily on direct retrieval from audio tokens by answer tokens, whereas AQA depends more strongly on a mediated route in which audio information is first integrated into prompt tokens and subsequently accessed during generation. We further probe prompt-token representations under audio concatenation and find that task-relevant audio attributes remain readily decodable, particularly in middle and later decoder layers, even when AQA performance deteriorates sharply. This dissociation indicates that the failure cannot be explained by complete loss of audio information from the decoder states and is instead consistent with a downstream bottleneck in retrieving or utilizing prompt-mediated information during answer generation. Together, our findings reveal task-dependent audio information routing in LALMs and highlight information utilization as a potential limitation on their generalization.

Omnimodal large language models (Omni-LLMs) integrate audio, video, and text, yet remain vulnerable to cross-modal hallucinations, where one modality improperly influences predictions about another. Existing training-free decoders modulate modality influence through perturbation or relevance weighting, but do not assess predictive compatibility within the joint audio-visual branch. Because joint-branch discrepancies may indicate either harmful interference or useful complementarity, reliable intervention requires assessing both discrepancy magnitude and actionability. To this end, we propose Conflict-Aware Decoding (CAD), a training-free framework comprising Potential Conflict Magnitude Estimation (PCME) and Conflict Actionability Assessment (CAA). PCME quantifies potential conflict using audio-video disagreement and the deviation of the joint prediction from a relevance-weighted unimodal reference. CAA then applies Dempster-Shafer reliability discounting to task-space answer relations, using query relevance and answer decisiveness to determine whether intervention is warranted. When an actionable conflict is identified, CAD selectively reallocates decoding weight from the joint branch to the unimodal branches. Experiments on CMM, AVHBench, WorldSense, and VideoMME show that CAD consistently outperforms the base decoder and competitive training-free methods across multiple audio-visual backbones. On Qwen2.5-Omni-7B, CAD improves overall accuracy by 14.1 and 8.0 percentage points on CMM and AVHBench, respectively, without model retraining.

Voice agents must call tools and hold multi-turn dialogue entirely through speech, yet the dominant paradigm trains them in text. Existing frameworks either cascade TTS and ASR around a proprietary voice API, where gradients cannot flow and per-call cost makes on-policy reinforcement learning prohibitive, or stay in text: they measure voice agents but cannot improve them. We present SpeechGym, an audio-native agentic environment in which two omni-modal models converse in native audio, with no external ASR or TTS and no API boundary, over the unmodified tasks, tools and success check of an established text agentic benchmark, so that the interaction modality is the only variable and the loop stays local and trainable end to end. Audio agentic capability does not follow from audio understanding. The failures speech introduces are perceptual rather than reasoning deficits: the agent picks the right tool and the right argument slot but fills it with a value misheard from the waveform, and that single error cascades into a failed call, a retry of the same call, and a wasted step budget. A second failure is behavioural: under an insistent caller the agent performs an unauthorised write and ends the episode believing it helped. Both are trainable, because the environment labels them for free: a call with a misheard argument fails against the database while a correct one succeeds. The obstacle is sparsity, not signal. Outcome-only GRPO is gradient-starved here, since almost every rollout group fails identically, while a per-turn process reward crediting each successful tool call restores variance to nearly every group. Trained this way, the agent transfers with no further tuning to an independently implemented voice benchmark, more than doubling task success and carrying an open-weights model from last place to second on that leaderboard, while using fewer turns and tokens than before training.

General audio comprehension now covers speech, sound, and music over durations from seconds to hours, driven by large audio-language models (LALMs) that are increasingly omni-modal. Yet the benchmarks that test them still rely on clips of seconds, where scores saturate and models converge; recent long-form efforts extend duration but evaluate long audio much as short clips are. We introduce LongAudioSpan, a benchmark that spans both duration and depth: it pairs audio from 10 minutes to over 2 hours with 3,240 questions across three cognitive levels, namely perception, understanding, and reasoning. Two paths supply the questions, differing in how question content is sourced and how ground truth is obtained. Native QA extracts questions from the audio's content, posing each as a multiple-choice item and an open-ended one graded by detailed rubrics. Anchor QA instead injects ground truth, planting acoustic anchors into the audio and building a perception-to-reasoning chain scored only to the first error. A fully automated pipeline constructs every item through structured captioning, QA generation, and adversarial critic feedback. Evaluating 12 LALMs on LongAudioSpan, we find the hard part comes before reasoning: distilling a few relevant facts from a long, redundant signal. This difficulty grows with audio length and falls hardest on perception, especially temporal grounding. LongAudioSpan is available at https://huggingface.co/datasets/holvan/LongAudioSpan.

We investigate whether AI-generated impulsive sounds can be distinguished from real ones through group delay analysis. Our central finding is that AI-generated impulsive sounds show near-identical onset-region group-delay distributions but exhibit measurably different group-delay behavior in the late decay region: decay-region KL divergence reaches $0.322$ compared to near-zero onset divergence ($0.022$). Cross-band GD variability achieves single-feature AUC~=~0.720, and a Random Forest (RF) over nine decay-region features reaches AUC~$=$~0.884 under sample-disjoint evaluation. A group delay map used as a standalone 2D input to CNN classifiers achieves 90–94% accuracy, demonstrating that group delay carries substantial discriminative information. Under generator hold-out, CNN and transformer classifiers show highly variable AUC (0.457–0.918). The group delay RF achieves the highest average hold-out accuracy among the evaluated methods ($66.7\%$) and avoids extreme below-random collapse, although its average AUC (0.731) is lower than CNN avg (0.762) and AST (0.772). Parameter sensitivity analysis across 27 STFT configurations confirms that the RF AUC remains stable (0.700–0.847, std~=~0.035). These results suggest that decay-region group delay can serve as a physically interpretable forensic cue that complements magnitude-based classifiers, while broader validation remains necessary.

Recent advancements in generative models are pushing video generation toward unbounded streaming audio-video generation for real-time interactive worlds. However, existing benchmarks primarily evaluate completed sequences and struggle to capture streaming properties. To bridge this gap, we introduce StreamAV-Bench, the first comprehensive benchmark tailored for streaming audio-video generation. StreamAV-Bench establishes a unified evaluation framework, including the progressive track for instruction adherence and long-horizon stability, and the interactive track for interactive response and state retention and reuse. With expert-verified evaluation cases across 32 fine-grained dimensions, we conduct an extensive evaluation of 13 representative systems. Our analysis reveals that current models suffer from temporal drift in progressive generation and responsiveness bottlenecks during interactive control. Based on a comprehensive failure analysis, we share insights to advance the development of native joint audio-video streaming models.

Large language model (LLM)-based audio-visual speech recognition (AVSR) systems are robust under noise. Contrastive decoding (CD), originally introduced to stabilize LLM generation by contrasting a weaker model against a stronger one at inference time, adjusts predictions without additional training. In this work, we apply CD to AVSR by contrasting audio-only conditioning with full audio-visual conditioning within the same underlying model. However, using a fixed contrastive strength introduces a trade-off across noise levels: stronger intervention helps under severe noise but may over-correct reliable predictions in clean conditions. We propose reliability-aware scaling of CD for AVSR. Instead of using a fixed strength, we adaptively modulate the contrastive influence at each token based on reliability signals derived from attention dynamics and inter-model predictive divergence. Experiments on LRS3 show consistent improvements across clean and low-SNR conditions.

Conversational systems, such as duplex speech language models (SLMs), still lack a streaming, accurate, and empathetic memory system as their soul. We introduce VoiceMem, a simple memory architecture with a parallel informational left brain, an emotional right brain, and streaming memory I/O mechanisms. We further build a complete pipeline for memory-aware SLM training, long-horizon evaluation, and decoupled deployment with interchangeable memory backends. Experiments and real-world deployment show three advantages: i) Accuracy: under top-5 retrieval, the left brain outperforms classical systems such as Mem0 at top-200 by nearly 30 points; ii) Emotional & Personal: the right brain, with short- and long-horizon affective attribution and dual-node persona modeling, achieves state-of-the-art performance across three persona benchmarks and improves the aggregate score by 4.29 points over the previous best system; and iii) Real-Time & Cheap: VoiceMem completes retrieval in 134 ms, well within standard VAD latency, adding no extra conversational delay while maintaining high accuracy and low cost. These results show that VoiceMem provides a practical memory foundation for real-time, personalized, and emotionally aware speech interaction.

Cough acoustics are promising for non-invasive tuberculosis (TB) screening, yet whether machine learning (ML) models capture disease-related acoustics or artifacts of data collection remains unresolved. We evaluated the cross-dataset generalizability of classical ML and deep learning (DL) cough-based TB classifiers across three independent datasets. Despite moderate within-dataset performance (ROC-AUC up to $0.755 \pm 0.056$), both pipelines fail to generalize, with external performance frequently below 0.6, indicating a possible limitation of the data. We further observed audio representations are organized by recording device and dataset rather than TB status, predicted TB probability tracks country-level prevalence in CODA, and device mismatch degrades transfer while device-diverse training improves it. Additionally, a clinical-variable baseline generalizes more consistently (ROC-AUC $0.655 - 0.711$), indicating acquisition-specific variability is a stronger driver of poor generalizability than population shift. High within-dataset performance is not enough. External validation is essential before cough-based TB models are clinically ready.