arXiv_SD
-
UniVocal: Unified Speech-Singing Code-Switching Synthesis
2026-06-01 04:35:28Yufei Shi, Qian Chen, Wen Wang, Xiangang Li, Zhen-Hua Ling, Yang AiAbstract
We propose UniVocal, a unified framework that implicitly infers vocal modes from text context to pioneer Speech-Singing Code-Switching (SCS) Synthesis - a task where transitions are autonomously driven by textual semantics, akin to seamless human language blending. Unlike single-mode generation or systems relying on switching-control tags, our proposed UniVocal implicitly infers vocal modes solely from text context. To achieve this, we employ a data-efficient two-stage curriculum learning strategy that progressively trains a competitive TTS system to acquire the desired SCS capability. Addressing data scarcity, we introduce a scalable pipeline to synthesize diverse code-switching data that is both semantically and acoustically natural, alongside a new multi-scenario benchmark, SCSBench. To address limitations of semantic tokenizers in capturing acoustic details, we also introduce refined cent token and Chain-of-Thought (CoT) generation for planning prosody before content generation, effectively enhancing empathetic speech generation and singing melody. Experimental results demonstrate that UniVocal achieves state-of-the-art performance on SCSBench while maintaining competitive performance on regular speech and singing tasks. Audio samples are available at this https URL. The code and dataset are released at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2606.01677
PDF
-
Description and Discussion on DCASE 2026 Challenge Task 2: Noise-aware Unsupervised Anomalous Sound Detection for Machine Condition Monitoring
2026-06-01 02:25:54Tomoya Nishida, Noboru Harada, Daiki Takeuchi, Daisuke Niizumi, Keisuke Imoto, Kota Dohi, Harsh Purohit, Takashi Endo, Yohei KawaguchiAbstract
This paper presents an overview of DCASE 2026 Challenge Task 2, titled "Noise-aware unsupervised anomalous sound detection (UASD) for machine condition monitoring." The task aims to advance noise-robust anomalous sound detection for machine condition monitoring under the unsupervised setting, where only normal machine sounds are available for training. Reliable detection under noisy conditions is crucial for practical deployment, but previous DCASE Task 2 settings provided limited information about environmental noise, potentially limiting UASD performance in highly noisy situations. To address this limitation, DCASE 2026 allows participants to exploit two-channel audio samples simultaneously captured at locations near and far from the target machine. Since the distant microphone is expected to contain relatively stronger environmental noise and weaker direct machine sounds, it may help distinguish environmental noise components from the target machine sounds. After the challenge submission deadline, challenge results and an analysis of the submitted systems will be added.
Abstract (translated)
URL
https://arxiv.org/abs/2606.01578
PDF
-
A Lightweight Slot-Attention Framework for Multi-Instrument Multi-Pitch Estimation
2026-05-31 21:42:46Michael TaenzerAbstract
Multi-pitch estimation (MPE) typically predicts which pitches are active in a mixture, but not which instrument or source produced them. This paper investigates a lightweight slot-attention framework for multi-instrument MPE (MI-MPE), where a mixture CQT is mapped to an unordered set of source-like pitch maps. The model uses permutation-invariant Hungarian matching to avoid fixed output semantics and treats the number of slots as an upper bound on the number of active sources. We further study two modular extensions: a self-supervised timbre encoder that provides training-time targets for slot-level timbre embeddings, and a polyphony branch that regularizes the pitch density of mixture- and slot-level predictions. Experiments show that Hungarian matching substantially improves instrument family decomposition on URMP. Stem-level prediction remains more challenging: timbre and polyphony supervision improve selected configurations, but do not consistently resolve source assignment. The results suggest that slot-based architectures are a promising direction for source-aware MPE, while highlighting the need to couple auxiliary musical cues to slot identity more carefully.
Abstract (translated)
URL
https://arxiv.org/abs/2606.01460
PDF
-
A 1000-hour EEG-EMG-audio dataset of Japanese speech production
2026-05-31 14:30:46Motoshige Sato, Ilya Horiguchi, Masakazu Inoue, Kenichi Tomeoka, Eri Hatakeyama, Yuya Kita, Atsushi Yamamoto, Ippei Fujisawa, Shuntaro SasaiAbstract
We present a multimodal dataset of 1020 hours of simultaneously recorded scalp electroencephalography (EEG), facial electromyography (EMG), and speech audio from three healthy native Japanese speakers during open-vocabulary overt speech. Recordings were acquired with three EEG systems-an ultra-high-density system (this http URL) and two cap-type systems (this http URL and eegosports), spanning 62-128 channels-across many sessions over several months. Each session provides time-synchronized EEG, facial EMG, and audio, together with speech-event annotations and transcriptions. Although collected with speech decoding as a primary motivation, the dataset also supports work on multimodal signal processing, artifact modeling, longitudinal and cross-device adaptation, and EEG representation learning. Technical validation included power spectral density and event-related potential analyses across participants, devices, and tasks, which showed the expected 1/f spectral profile, task-related alpha-band attenuation, and time-locked evoked responses. The dataset is released in Brain Imaging Data Structure (BIDS) format via OpenNeuro under a CC0 waiver to support both speech-related and broader EEG research.
Abstract (translated)
URL
https://arxiv.org/abs/2606.01264
PDF
-
Spiking and Event-driven Neuromorphic Mamba Models for Efficient Speech Recognition
2026-05-31 10:15:08Tauseef Ahmed, Tao Sun, Jeronimo Castrillon, Kanishkan Vadivel, Guangzhi TangAbstract
Deep learning has greatly advanced automatic speech recognition (ASR), enabling widespread deployment on edge devices such as smartphones and smart home systems. However, the computational and energy demands of deep neural networks pose significant challenges for such resource-constrained deployments, introducing latency and limiting real-time interaction. Neuromorphic computing offers a promising solution by introducing activation sparsity through spiking neural networks (SNNs) and event-driven neural networks, converting dense operations into sparse computations. However, a study that evaluates the hardware benefits of different neuromorphic strategies remains lacking for ASR. This paper explores spiking and event-driven neuromorphic neural networks to improve activation sparsity in the state-of-the-art SpeechMamba model for ASR. We introduce an event-driven SpeechMamba with FATReLU activation, achieving over 60% activation sparsity with less than 1% accuracy degradation on LibriSpeech. We also propose a spiking SpeechMamba that attains over 70% sparsity while using 30% fewer parameters than comparable SNNs. Finally, we develop a cycle-accurate event-driven simulator enabling flexible algorithm-hardware co-exploration, which helps us identify computational bottlenecks and yields over 10% additional efficiency improvements.
Abstract (translated)
URL
https://arxiv.org/abs/2606.01135
PDF
-
Context-aware child-directed speech detection from long-form recordings
2026-05-31 10:12:47Th\'eo Charlot, Tarek Kunze, Kaveri K. Sheth, Alejandrina Cristia, Marvin LavechinAbstract
Automatically distinguishing child-directed speech from adult-directed speech in long-form recordings is key to scalable analyses of children's language environments. Existing approaches process utterances in isolation and have been evaluated primarily on English. We address these gaps along three dimensions. First, we fine-tune and evaluate six-self supervised models on a multilingual dataset of 182 children, showing that in-domain pre-training on child-centered recordings substantially outperforms models trained on adult speech. Second, we demonstrate that incorporating surrounding context substantially improves classification, with an absolute gain of 13.8% in average F1-score. Third, we evaluate our model in a realistic end-to-end pipeline, from adult speech detection to addressee classification, showing that performance drops under automatic segmentation but still consistently outperforms a rule-based baseline.
Abstract (translated)
URL
https://arxiv.org/abs/2606.01134
PDF
-
MelT: GEMM-Native NDFT for Efficient Single-Stage Audio Frontends on Modern Accelerators
2026-05-31 04:53:12Augusto Camargo, Marcelo FingerAbstract
Modern audio processing networks are commonly deployed on accelerators whose peak throughput is obtained through dense linear algebra, whereas conventional acoustic frontends -- a Short-Time Fourier Transform (STFT) followed by sparse Mel aggregation -- remain structurally heterogeneous. This mismatch can introduce memory-bandwidth, dispatch, and intermediate-allocation overheads on contemporary accelerator backends. This work introduces MelT, a single-stage frontend framework in which Mel-spaced Non-Uniform Discrete Fourier Transform (NDFT) bases are precomputed and applied to time-domain acoustic frames through dense General Matrix Multiplication (GEMM) operations. The contribution is not the NDFT operator itself; rather, it is the formulation of Mel-spaced NDFT projection as a GEMM-native audio frontend and its evaluation as a hardware-efficient alternative to conventional STFT+Mel pipelines. Evaluated across platforms ranging from Apple A18 Pro edge hardware to NVIDIA H100 datacenter acceleration, MelT attains up to a $3.75\times$ speedup in inference latency and a $3.52\times$ reduction in energy consumption while maintaining downstream classification accuracy.
Abstract (translated)
URL
https://arxiv.org/abs/2606.01009
PDF
-
Quality Audio Prototyping: a prototype system for unified sound retrieval and procedural generation
2026-05-30 09:05:11Nelly Garcia, Aditya Bhattacharjee, Gabryel Mason-Williams, Israel Mason-Williams, Emmanouil Benetos, Joshua ReissAbstract
Sound design workflows frequently oscillate between time-consuming library searches and the complexity of procedural synthesis, with practitioners typically relying on disconnected tools to address each challenge separately. This paper introduces Quality Audio Prototyping (QuAP), a working prototype that unifies content-based audio retrieval and procedural sound generation within a single interface, reducing the procedural distance between a narrative concept and its sonic realisation. QuAP integrates a similarity-based retrieval engine with real-time procedural audio models, complemented by a rule-based assistant that provides perceptually informed parameter guidance, offering definitions and recommendations derived from empirical optimisation rather than requiring prior synthesis knowledge. Preliminary evaluation confirms the viability of this approach: subjective assessment demonstrated statistically significant quality improvements in five of six embedded synthesis models, and an encoder ablation study established the preferred retrieval architecture on a sound effect dataset. A user evaluation with 16 practitioners confirmed the tool's workflow utility, with all participants agreeing that the parameter assistant preserved creative agency while lowering the barrier to procedural interaction.
Abstract (translated)
URL
https://arxiv.org/abs/2606.00629
PDF
-
UNISON: A Unified Sound Generation and Editing Framework via Deep LLM Fusion
2026-05-29 16:43:07Zhaoqing Li, Haoning Xu, Jingran Su, Yaofang Liu, Zhefan Rao, Huimeng Wang, Jiajun Deng, Tianzi Wang, Zengrui Jin, Rui Liu, Haoxuan Che, Xunying LiuAbstract
We present UNISON, a latent diffusion framework that unifies speech generation, sound generation, and audio editing within a single model. A single model handles text-to-audio, text-to-speech, zero-shot speaker cloning, mixed speech-and-sound generation, scene-level audio editing, speech-in-scene editing, and timed temporal composition, all of which share a single set of weights. Our architecture features two core designs: (1) Layer-wise deep LLM fusion, which injects hidden states from uniformly sampled layers of a frozen MLLM into corresponding MM-DiT blocks via learned projections, providing depth-matched semantic conditioning that improves instruction following over single-layer baselines; and (2) a unified multi-task architecture where task identity is encoded solely by a channel-wise mask and source audio is provided through VAE-encoded channel concatenation. Training is stabilized by an online GPU-side multi-task data synthesis pipeline with task-homogeneous batching and a two-stage curriculum. With 621M--732M trainable parameters, UNISON achieves results competitive with or exceeding task-specialist models across evaluated domains, while being roughly $4\times$ smaller than comparable unified systems.
Abstract (translated)
URL
https://arxiv.org/abs/2605.31530
PDF
-
Subtitle-Aligned Fine-Tuning of Whisper for Swiss German ASR: Benchmark Contamination, Convention Mismatch, and an Honest Baseline at 25.6% WER
Abstract
We present a systematic study of fine-tuning OpenAI's Whisper large-v3 for Swiss German ASR, using 1,367 hours of broadcast speech paired with Standard German subtitles as weak supervision. Through 16 iterative training runs on an NVIDIA DGX Spark (Grace Blackwell, 128 GB unified memory, up to 1 PFLOP FP4), we compare LoRA and full fine-tuning of the 1.55B-parameter model, investigate hallucination root causes, and quantify the effect of data quality, subtitle alignment, and training strategy. Our best model achieves 25.6% measured WER on the All Swiss German Dialects Test Set (ASGDTS) in an honest evaluation on strictly disjoint data. A harmonized error analysis separating genuine errors from valid stylistic variation (tense, word order, Swiss orthography) yields a content WER (cWER) of 13.8%, counting only actual recognition failures. Bias-corrected estimation reduces this to 8.5%, suggesting the true error rate is roughly one third of measured WER. We demonstrate that published state-of-the-art Swiss German ASR results (17.1-17.5% WER) are inflated by benchmark contamination: a vanilla Whisper model self-trained on the ASGDTS test set with zero Swiss German data achieves 13.88% WER, surpassing all published systems. Experiments with Phi-4-multimodal show an even stronger memorization effect (3.9% WER), revealing that the benchmark primarily measures convention matching rather than dialectal comprehension. We release two models, a LoRA adapter (25.32% WER, 13.9% cWER) and a full fine-tuned model (25.60% WER, 13.8% cWER), among the few publicly available, honestly evaluated Whisper models for Swiss German, under Apache 2.0 with full reproducibility, requiring no institutional data agreements.
Abstract (translated)
URL
https://arxiv.org/abs/2606.07608
PDF
-
Sound effects in media:A comparative analysis of recorded and synthetic samples in live-action and animation
Abstract
Creating sound for storytelling is crucial to establishing the environment in productions such as films, TV series and video games. This process often involves repeating, layering and recording real objects or using sound libraries, which can be time-consuming and repetitive. To address these challenges, procedural audio, also known as digital foley, offers a solution by allowing sound designers to quickly generate samples. Despite its efficiency, questions remain about the believability of synthetic samples compared to real ones. In our study, we compared synthetic samples generated by an online procedural engine and integrated them with both animated and live-action visuals. Our results indicate that procedural audio is highly effective and perceived as believable in drama and sci-fi scenes, particularly for sound models such as lasers, hits, air and rockets, whereas synthetic sounds weren't as believable in cartoon productions when representing everyday actions. Finally, we identified specific models that needed optimisation and highlighted audio features that needed improvement with feedback from audio professionals.
Abstract (translated)
URL
https://arxiv.org/abs/2605.31082
PDF
-
Towards Streaming Synchronized Spatial Audio Generation via Autoregressive Diffusion Transformer
2026-05-29 07:28:24Ke Lei, Yu Zhang, Changhao Pan, Xueyi Pu, Wenxiang Guo, Ruiqi Li, Zhou ZhaoAbstract
Real-time and accurate spatial audio generation is pivotal for delivering an immersive experience. However, existing spatial audio synthesis technologies are often encumbered by a tradeoff between generation quality and high inference latency, as well as difficulty in capturing precise spatial information from multimodal inputs. To address these challenges, we propose SwanSphere, a unified streaming framework for high-fidelity spatial audio generation from panoramic videos and text prompts. SwanSphere mainly makes the following contributions: 1) We introduce a causal autoregressive diffusion transformer architecture that enables streaming high-quality spatial audio generation. 2) We design a Spatial Video-Audio Contrastive (SVAC) learning strategy to align the video encoder with the acoustic domain, and further employ a multi-objective online direct preference optimization (ODPO) scheme, resulting in strong spatial perception and robust multimodal spatial audio synthesis. 3) To alleviate the current scarcity of spatial audio datasets, we also develop an automated annotation pipeline for generating detailed spatial captions. Experimental results demonstrate that SwanSphere achieves superior performance in both video-to-spatial and text-to-spatial audio generation tasks. Demos can be found at: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2605.30940
PDF
-
Audio Pirates: Black-box Audio Watermark Removal via Diffusion Priors
2026-05-28 22:07:32Lingfeng Yao, Xincong Zhong, Chenpei Huang, Xuandong Zhao, Hanqing Guo, Aohan Li, Jiang Liu, Tomoaki Ohtsuki, Miao PanAbstract
With the rise of AI-generated audio, watermarking has become widely used for detecting misuse and protecting intellectual property. However, adversaries may try to remove these watermarks, making it critical to evaluate how well watermarking schemes withstand removal attacks. Existing attacks are often impractical: they either noticeably degrade perceptual quality or require access to the watermarking scheme. We propose DiffErase, a black-box watermark removal attack that assumes no knowledge of the target watermarking scheme while maintaining perceptual quality. DiffErase perturbs watermarked audio to an intermediate diffusion noise level and regenerates it using a pretrained denoising model, effectively suppressing watermark signals. Theoretical analysis and extensive experiments demonstrate that inaudible audio watermarks are highly vulnerable: across multiple audio domains, DiffErase consistently removes watermarks while preserving perceptual quality. These findings highlight the need for future audio watermarking designs to consider diffusion-based threats. Code and demos are available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2605.30614
PDF
-
Decoding Strategies for Diffusion-Based ASR: A Systematic Evaluation of Confidence-Based Thresholding
2026-05-28 08:48:56Jeong Hun Yeo, Minsu Kim, Hyeongseop Rha, Yong Man RoAbstract
While LLM-based Automatic Speech Recognition (ASR) achieves high accuracy, its speed is limited by sequential autoregressive decoding. Diffusion Language Models (DLMs) offer a parallel alternative, yet their decoding strategies remain under-explored in ASR contexts. This paper analyzes three decoding schemes for DLM-based ASR: fixed-number, static confidence threshold, and dynamic confidence threshold. We propose measuring round-wise accuracy using Negative Log-Likelihood-based uncertainty as a proxy for decoding progress. Our results show that both threshold-based strategies significantly outperform fixed-number schemes in accuracy and speed. We attribute this to a property unique to ASR: most tokens reach high confidence early, allowing reliable ones to be harvested aggressively while leaving only difficult tokens for later rounds. Notably, the static-threshold strategy matches the accuracy of autoregressive decoding while offering superior efficiency.
Abstract (translated)
URL
https://arxiv.org/abs/2605.29613
PDF
-
ChildVox: A Speech, Audio, and Large Audio-Language Model Benchmark in Understanding and Characterizing Sound across Childhood
2026-05-28 02:22:20Tiantian Feng, Anfeng Xu, Xuan Shi, Aditya Kommineni, Shakhrul Iman Siam, Megan Micheletti, Zhonghao Shi, Helen Tager-Flusberg, Mi Zhang, Lynn K. Perry, Catherine Lord, Daniel Messinger, Shrikanth NarayananAbstract
We present ChildVox, a novel benchmark for characterizing the diverse acoustic signals through which children communicate. Specifically, ChildVox follows the full developmental trajectory from birth through school age, covering physiological sounds, non-linguistic vocalizations, canonical syllables, and spoken language. ChildVox integrates more than 20 sub-tasks across 17 child-centered audio and speech datasets, enabling systematic cross-corpus and cross-domain comparison. We evaluate a representative range of audio and speech foundation models, including self-supervised, ASR-oriented, and large audio-language models, on tasks including physiological sound classification, vocalization and canonical syllables modeling, and speech quality assessment and recognition. Benchmark results show that ChildVox provides a suite of high-performance models in recognizing a wide range of acoustic signals from children, supporting downstream applications such as characterizing children's language levels and tracking speech production with age.
Abstract (translated)
URL
https://arxiv.org/abs/2605.29257
PDF
-
Affective Music Recommendation: A Rollout-Based World Model for Offline Preference Optimization
2026-05-27 17:58:46Audrey Chan, Aaron Labb\'e, Jacob Lavoie, Jordan Bannister, Ars\`ene Fansi Tchango, Guillaume Lajoie, Laurent CharlinAbstract
Functional music applications, from consumer focus and sleep aids to clinical interventions, share a distinctive recommendation problem: success is defined by the listener's affective state, but online experimentation on emotion is ethically constrained, particularly for clinical populations who cannot reliably skip a song or report distress. We describe AMRS, the Affective Music Recommendation System deployed on LUCID's health-and-wellness platforms, which serve clinical users (primarily older adults with neurocognitive conditions) and consumer-wellness users across energize, focus, calm, and sleep modes. AMRS is built around a rollout-based world model: a causal transformer trained on logged listening data to jointly predict engagement, binary rating, and self-reported valence and arousal. The world model serves both as an in-silico simulator for offline policy training and as a stress-testing tool before deployment. A recommender policy initialized by behaviour cloning is fine-tuned offline with Direct Preference Optimization (DPO) against a configurable multi-objective utility function. Under a strict cold-start protocol, the world model predicts both behavioural and affective signals with usable fidelity; DPO improves predicted valence and arousal over the cloned baseline while maintaining a similar diversity profile and avoiding the distributional collapse produced by greedy optimization. We position the work as an early deployed validation of a methodology for affective recommendation when online experimentation is ethically untenable.
Abstract (translated)
URL
https://arxiv.org/abs/2605.28810
PDF
-
Cross-modal characterization of infant cry: validation of a chest-surface accelerometer in extracting acoustic vocal function measures
2026-05-27 16:17:25Winko W. An, Saketh Sundar, Lisa Yankowitz, Daryush D. Mehta, Carol L. WilkinsonAbstract
Background: Infant cry acoustics provide a promising window into early neurodevelopment and may serve as scalable biomarkers for neurodevelopmental disorders. However, conventional microphone-based recordings are highly susceptible to environmental noise and raise privacy concerns in real-world clinical settings. Chest-surface accelerometers may offer a robust alternative by capturing vibrations directly from the larynx. Methods: We evaluated the validity of a chest-mounted accelerometer (ACC) for infant cry analysis by comparing acoustic features derived from ACC and simultaneously recorded microphone (MIC) signals during routine vaccination visits. The final sample included 85 infants (41 at 4 months; 44 at 12 months) from a diverse pediatric population. Seven vocal measures were extracted from both modalities, including fundamental frequency (F0), jitter, shimmer, cepstral peak prominence (CPP), and harmonics-to-noise ratio (HNR). Agreement and consistency between modalities was assessed using intraclass correlation coefficients (ICCs). Results: F0 demonstrated excellent agreement between ACC and MIC recordings (ICC > 0.94). Jitter measures also showed good-to-excellent agreement, while CPP demonstrated moderate agreement. Shimmer and HNR showed lower absolute agreement and systematic bias between modalities, reflecting possible differences in signal transmission and noise sensitivity. Conclusion: In summary, chest-surface accelerometers can reliably capture several clinically relevant acoustic features of infant cry, particularly temporal measures of F0 and jitter. This approach offers a noise-robust and privacy-preserving alternative to microphone-based recordings, supporting its potential use in scalable clinical and developmental research applications.
Abstract (translated)
URL
https://arxiv.org/abs/2605.28687
PDF
-
DEMON: Diffusion Engine for Musical Orchestrated Noise
Abstract
We present DEMON, a real-time diffusion engine that makes the denoising process playable as a live musical instrument: a control surface both broad (many parameters shaped per-frame across the output) and responsive (each control taking effect as fast as its place in the denoising loop allows). Built on ACE-Step 1.5 and StreamDiffusion's ring-buffer architecture with TensorRT acceleration, it sustains up to 12.3 decoder completions per second for 60-second music on a single consumer GPU (RTX 5090), or 11.3 generations per second at our production ring-depth of 4. At these rates denoising parameters become viable as live performance controls, but the ring buffer propagates per-request changes only at its drain rate, a floor of S denoising steps. We contribute four mechanisms. (1) Per-slot heterogeneous denoise scheduling: each ring-buffer slot owns its timestep schedule, so a moving denoise slider is tracked without wiping the in-flight queue, where the upstream global-schedule design must rebuild and discard it. (2) Shared mutable per-step state, giving any parameter consulted at every solver step next-tick effect, bypassing ring-buffer drain. (3) Per-frame source blending: a sampling-time control on the standard SDE re-noise step, giving a framewise transformation-strength axis that complements scalar denoise scheduling. (4) Windowed VAE decode exploiting receptive-field analysis for an 8.0x decode speedup. Together these separate streaming-diffusion parameters into four propagation classes, by onset and convergence latency.
Abstract (translated)
URL
https://arxiv.org/abs/2605.28657
PDF
-
Audio-Mind: An Auditable Agentic Framework for Audio Understanding
2026-05-27 13:39:14Yucheng Wang, Jing Peng, Hanqi Li, Chenghao Wang, Wenming Tu, Yu Xi, Zhaokai Sun, Kai Yu, Shuai WangAbstract
Audio agents extend large audio-language models (LALMs) by decomposing audio questions into tool calls, intermediate evidence, and iterative reasoning steps. However, as LALMs become stronger, the key challenge shifts from enabling tool use to determining when agentic evidence acquisition genuinely benefits audio understanding. We propose Audio-Mind, an auditable and pluggable framework for conditional evidence acquisition in audio understanding. Audio-Mind dynamically combines a strong frontend with planner-guided tool use, preserving frontend judgment when initial evidence is sufficient while acquiring bounded external evidence for questions with unresolved evidence gaps. Experiments on MMAR and MSU-Bench show that Audio-Mind outperforms prior audio-agent baselines, reaching 80.4% accuracy on MMAR and 82.8% accuracy on MSU-Bench. A matched-backbone comparison highlights why this design matters: under strong audio frontends, agentic decomposition can become an orchestration bottleneck when the workflow does not preserve the frontend's holistic audio-grounded judgment. Beyond accuracy, Audio-Mind produces higher-quality, auditable reasoning traces that expose uncertainty, tool evidence, and answer rationales, offering a potential basis for more reliable audio-QA annotation and error analysis.
Abstract (translated)
URL
https://arxiv.org/abs/2605.28480
PDF
-
VoiceGiraffe: A Benchmark for Extreme Long-Context Audio-Language Understanding
2026-05-27 05:15:14Jashin Ye, Dongxiao Wang, Yixuan Ye, Sashuai Zhou, Weihuang Lin, Mingyang Han, Kunpeng Wang, Zeyu Yuan, Boyu Li, Haoxiang Shi, Jingchen Shu, Jun Song, Bo ZhengAbstract
While large audio language models (LALMs) have achieved remarkable progress in audio processing at the second- or minute-level scale, understanding hour-level audio remains a fundamental bottleneck. Existing benchmarks predominantly rely on short clips or artificially concatenated segments, failing to faithfully assess LALM capacity for long-range information comprehension in real-world scenarios such as podcasts and lengthy speeches. To address this gap, we introduce VoiceGiraffe, a novel benchmark designed to rigorously evaluate LALMs across diverse real-world scenarios, modalities, and languages under long-context settings. It comprises 1500 curated triplets structured into a dual-level taxonomy of single-hop perception and multi-hop reasoning. We evaluate a broad suite of open-source and proprietary LALMs against human performance. Results underscore three fundamental findings. First, VoiceGiraffe remains highly challenging and far from saturation. Second, we show that no single inference paradigm universally dominates. The E2E inference benefits models with native long-context audio understanding, cascaded caption aggregation stabilizes small models overwhelmed by hour-scale audio, and reasoning-enhanced cascading with external LLM helps weaker models but can bottleneck stronger proprietary systems. Third, we reveal long-range memory persistence as a key bottleneck. LALMs are better at answering questions that require connecting salient causal cues than those requiring sustained tracking of sparse events across long audio, whereas humans show the opposite pattern. These findings position VoiceGiraffe as a challenging and diagnostic testbed for long-form audio understanding, highlighting the need for LALMs with persistent memory and robust long-range aggregation.
Abstract (translated)
URL
https://arxiv.org/abs/2605.27976
PDF
