arXiv_SD
-
JASTIN: Aligning LLMs for Zero-Shot Audio and Speech Evaluation via Natural Language Instructions
2026-05-06 05:18:42Leying Zhang, Bowen Shi, Haibin Wu, Bach Viet Do, Yanmin QianAbstract
The rapid advancement of generative audio models has outpaced the development of robust evaluation methodologies. Existing objective metrics and general multimodal large language models (MLLMs) often struggle with domain generalization, zero-shot capabilities, and instructional flexibility. To address these bottlenecks, we propose JASTIN, a generalizable, instruction-driven audio evaluation framework that formulates audio assessment as a self-instructed reasoning task. JASTIN bridges a frozen high-performance audio encoder with a fine-tuned LLM backbone via a trainable audio adapter. To ensure robust zero-shot generalization, we introduce a comprehensive instruction following data preparation pipeline, incorporating Multi-Source, Multi-Task, Multi-Calibration, and Multi-Description data. Experimental results demonstrate that JASTIN achieves state-of-the-art Pearson and Spearman correlations with human subjective ratings. It consistently outperforms general MLLMs across speech, sound, music, and out-of-domain evaluation tasks without the need for task-specific retraining.
Abstract (translated)
URL
https://arxiv.org/abs/2605.04505
PDF
-
Adaptive Diagonal Loading for Norm Constrained Beamforming
2026-05-05 23:00:06Manan Mittal, Ryan M. Corey, John R. Buck, Andrew C. SingerAbstract
Reliable adaptive beamforming is critical for large microphone arrays operating in highly dynamic acoustic environments. In scenarios characterized by fast-moving talkers and interferers, the available sample support for estimating the spatial correlation matrix is often snapshot-deficient. This deficiency, coupled with array imperfections, degrades the White Noise Gain (WNG), leading to severe target signal cancellation. To ensure stable and robust beamforming, we propose a novel adaptive diagonal loading method that guarantees the WNG remains strictly within specified bounds. By leveraging the Kantorovich inequality, we map the desired WNG to a strict upper bound on the condition number of the correlation matrix. Furthermore, we present three estimation techniques for the adaptive loading level, ranging from trace-based bounding to exact eigenvalue decomposition, offering scalable computational complexities of $\mathcal{O}(M)$, $\mathcal{O}(M^2)$, and $\mathcal{O}(M^3)$. Our approach demonstrates highly stable beamforming under fast-changing interference.
Abstract (translated)
URL
https://arxiv.org/abs/2605.04342
PDF
-
ShipEcho -- An Interactive Tool for Global Mapping of Underwater Radiated Noise from Vessels
2026-05-05 17:59:18Mark Shipton, Valentino Denona, {\DJ}ula Na{\dj}, Roee DiamantAbstract
Underwater radiated noise from vessels (V-URN) is a recognized environmental stressor that negatively impacts marine ecosystems. Significant resources are invested in the development of V-URN monitoring indicators, regulatory frameworks, and management-oriented assessments. One approach with high potential for impact is V-URN mapping, which can provide actionable spatiotemporal information for environmental assessment and mitigation planning. Producing management-scale maps remains challenging as passive acoustic measurements are spatially sparse and many operational systems depend on specialist workflows and costly access to wide-area vessel activity data. To address these constraints, we introduce ShipEcho, a freely accessible web-based Geographic Information System (GIS) that provides near-real-time V-URN mapping using vessel data acquired through a community-based AIS exchange. Using established vessel SL models and propagation modeling informed by bathymetric data, ShipEcho produces near-real-time and cumulative noise maps across regions worldwide. These include sound pressure levels and sound exposure levels using standard indicators, including the 63~Hz and 125~Hz one-third octave bands and a 20--2000~Hz broadband level. We describe the system architecture, data pipeline, modeling workflow, and key assumptions, and evaluate map accuracy through comparison with acoustic recordings. We then demonstrate how ShipEcho can support management-level assessment, decision-making, and policy initiatives through practical use cases.
Abstract (translated)
URL
https://arxiv.org/abs/2605.08194
PDF
-
MiniMind-O Technical Report: An Open Small-Scale Speech-Native Omni Model
2026-05-05 16:27:33Jingyao GongAbstract
MiniMind-O is an open 0.1B-scale omni model built on the MiniMind language model. It accepts text, speech, and image inputs, and returns both text and streaming speech. The release includes model code, checkpoints, and the main Parquet training datasets for text-to-audio, image-to-text, and audio-to-audio training, making the complete interaction loop directly inspectable. The model uses a full MiniMind backbone as the Thinker and an independent four-layer Talker made from MiniMind blocks. Frozen SenseVoice-Small and SigLIP2 encoders provide speech and image features, which are mapped by lightweight MLP projectors and injected at modality-placeholder positions. The Talker reads a middle-layer Thinker state together with an autoregressive eight-layer Mimi-code buffer. Speaker control is handled by a dedicated speaker token, right-aligned reference codec prompts, and precomputed CAM++ speaker embeddings, so voice conditioning remains part of the audio-code context rather than a separate TTS module. With a 768-dimensional Talker, the dense and MoE variants reach average CERs of 0.0897 and 0.0900 in Thinker--Talker consistency evaluation, with overall voice-cloning similarities of 0.5995 and 0.5937. Beyond reporting a working system, the paper identifies three scale-critical design choices for small omni models: middle-layer semantic bridging, a released multimodal sequence format, and a parameter-efficient eight-codebook interface.
Abstract (translated)
URL
https://arxiv.org/abs/2605.03937
PDF
-
Towards Open World Sound Event Detection
2026-05-05 16:23:06P. H. Hai, L. T. Minh, L. H. SonAbstract
Sound Event Detection (SED) plays a vital role in audio understanding, with applications in surveillance, smart cities, healthcare, and multimedia indexing. However, conventional SED systems operate under a closed-world assumption, limiting their effectiveness in real-world environments where novel acoustic events frequently emerge. Inspired by the success of open-world learning in computer vision, we introduce the Open-World Sound Event Detection (OW-SED) paradigm, where models must detect known events, identify unseen ones, and incrementally learn from them. To tackle the unique challenges of OW-SED, such as overlapping and ambiguous events, we propose a 1D Deformable architecture that leverages deformable attention to adaptively focus on salient temporal regions. Furthermore, we design a novel Open-World Deformable Sound Event Detection Transformer (WOOT) framework incorporating feature disentanglement to separate class-specific and class-agnostic representations, together with a one-to-many matching strategy and a diversity loss to enhance representation diversity. Experimental results demonstrate that our method achieves marginally superior performance compared to existing leading techniques in closed-world settings and significantly improves over existing baselines in open-world scenarios.
Abstract (translated)
URL
https://arxiv.org/abs/2605.03934
PDF
-
PHALAR: Phasors for Learned Musical Audio Representations
2026-05-05 16:19:58Davide Marincione, Michele Mancusi, Giorgio Strano, Luca Cerovaz, Donato Crisostomi, Roberto Ribuoli, Emanuele Rodol\`aAbstract
Stem retrieval, the task of matching missing stems to a given audio submix, is a key challenge currently limited by models that discard temporal information. We introduce PHALAR, a contrastive framework achieving a relative accuracy increase of up to $\approx 70\%$ over the state-of-the-art while requiring $<50\%$ of the parameters and a 7$\times$ training speedup. By utilizing a Learned Spectral Pooling layer and a complex-valued head, PHALAR enforces pitch-equivariant and phase-equivariant biases. PHALAR establishes new retrieval state-of-the-art across MoisesDB, Slakh, and ChocoChorales, correlating significantly higher with human coherence judgment than semantic baselines. Finally, zero-shot beat tracking and linear chord probing confirm that PHALAR captures robust musical structures beyond the retrieval task.
Abstract (translated)
URL
https://arxiv.org/abs/2605.03929
PDF
-
Ecologically-Constrained Task Arithmetic for Multi-Taxa Bioacoustic Classifiers Without Shared Data
2026-05-05 16:10:02Ragib Amin Nihal, Benjamin Yen, Runwu Shi, Takeshi Ashizawa, Kazuhiro NakadaiAbstract
Training data for bioacoustics is scattered across taxa, regions, and institutions. Centralizing it all is often infeasible. We show that independently fine-tuned BEATs encoders can be composed into a unified 661-species classifier via task vector arithmetic without sharing data. We find that bioacoustic task vectors are near-orthogonal (cosine 0.01-0.09). Their separation aligns closely with spectral distribution distance, a gradient consistent with the acoustic niche hypothesis. This geometry makes simple averaging optimal while sign-conflict methods reduce accuracy by one to six percentage points. Composition also creates an asymmetric gap: species-rich groups lose accuracy relative to joint training while underrepresented taxa gain, a redistribution useful for equitable biodiversity monitoring. We verify linear mode connectivity across all taxonomic pairs, demonstrate zero-shot transfer to new regions, and identify domain negation as a boundary condition where composition fails. These results enable a collaborative paradigm for bioacoustics where institutions share only task vectors to assemble multi-taxa classifiers, preserving data privacy.
Abstract (translated)
URL
https://arxiv.org/abs/2605.03914
PDF
-
Cosmodoit: A Python Package for Adaptive, Efficient Pipelining of Feature Extraction from Performed Music
Abstract
Computational analysis of performed music is a key component of music information research, as performance shapes much of the music we hear. Music performance analysis studies the acoustic variations introduced by performers and how these variations reflect musical interpretation and structure. Although many algorithms and tools exist for tasks such as performance-to-score alignment and symbolic or audio feature extraction, they are spread across different programming languages and data formats, making them difficult to combine efficiently. To address this problem, we present Cosmodoit, a novel Python package designed to streamline feature extraction from performed music. Cosmodoit integrates performance-to-score alignment with symbolic and audio feature extraction in a modular, flexible pipeline that supports selective processing, dependency-aware computation, and incremental updates. Its extensible design reduces duplicated work, minimizes errors, and enables efficient large-scale processing. By accommodating algorithms implemented in multiple languages and allowing parameter tuning for consistent feature extraction, Cosmodoit provides a versatile and practical tool for both research and development in music performance analysis.
Abstract (translated)
URL
https://arxiv.org/abs/2605.03541
PDF
-
Deepfake Audio Detection Using Self-supervised Fusion Representations
2026-05-05 06:51:41Khalid Zaman, Qixuan Huang, Muhammad Uzair, Masashi UnokiAbstract
This paper describes a submission to the Environment-Aware Speech and Sound Deepfake Detection Challenge (ESDD2) 2026, which addresses component-level deepfake detection using the CompSpoofV2 dataset, where speech and environmental sounds may be independently manipulated. To address this challenge, a dual-branch deepfake detection framework is proposed to jointly model speech and environmental contextual representations from input audio. Two pretrained models, XLS-R for speech and BEATs for environmental sound, are used to extract complementary contextual representations. A Matching Head is introduced to model representation differences through statistical normalization and representation interaction, enabling estimation of the original class. In parallel, multi-head cross-attention enables effective information exchange between speech and environmental components. The refined representations are processed with residual connections and layer normalization, and passed to an AASIST classifier to predict speech-based and environment-based spoofing probabilities. The model outputs original, speech, and environment predictions. On the test set, the proposed system achieves an F1-score of 70.20% and an environmental EER of 16.54%, outperforming the baseline system.
Abstract (translated)
URL
https://arxiv.org/abs/2605.03420
PDF
-
Smart Passive Acoustic Monitoring: Embedding a Classifier on AudioMoth Microcontroller
2026-05-05 06:37:35Louis Lerbourg, Paul Peyret, Juliette Linossier, Marielle MalfanteAbstract
Passive Acoustic Monitoring (PAM) is an efficient and non-invasive method for surveying ecosystems at a reduced cost. Typically, autonomous recorders allow the acquisition of vast bioacoustic datasets which are then analyzed. However, power consumption and data storage are both scarce and limit the duration of acquisition campaigns. To address this issue, we propose a smart PAM system which allows the in-situ analysis of the soundscape by embedding a classifier directly onto an AudioMoth microcontroller. Specifically, we propose an optimized yet simple 1D Convolutional Neural Network (1D-CNN) to classify the raw audio. The model focuses on the specific call of Scopoli Shearwater seabirds (endangered species) and is trained on a real-world dataset with a classification accuracy of 91\% (balanced accuracy of 89\%). We also propose a process to optimize the model to fit the severe resource constraints of the AudioMoth, achieving a \~10kB RAM memory footprint and 20ms inference time. Finally, we present an open-source tutorial of our model optimization and export strategy which can be used for embedding models beyond the scope of our study. Our modified version of the AudioMoth firmware adds two functions: (F1) which selectively records data when the target species has been detected and (F2) which logs the continuous classification results in real time. This work intends to facilitate the conception of intelligent sensors, enhancing the efficiency and scalability of bioacoustic monitoring campaigns.
Abstract (translated)
URL
https://arxiv.org/abs/2605.03412
PDF
-
APEX: Large-scale Multi-task Aesthetic-Informed Popularity Prediction for AI-Generated Music
2026-05-05 06:11:51Jaavid Aktar Husain, Dorien HerremansAbstract
Music popularity prediction has attracted growing research interest, with relevance to artists, platforms, and recommendation systems. However, the explosive rise of AI-generated music platforms has created an entirely new and largely unexplored landscape, where a surge of songs is produced and consumed daily without the traditional markers of artist reputation or label backing. Key, yet unexplored in this pursuit is aesthetic quality. We propose APEX, the first large-scale multi-task learning framework for AI-generated music, trained on over 211k songs (10k hours of audio) from Suno and Udio, that jointly predicts engagement-based popularity signals - streams and likes scores - alongside five perceptual aesthetic quality dimensions from frozen audio embeddings extracted from MERT, a self-supervised music understanding model. Aesthetic quality and popularity capture complementary aspects of music that together prove valuable: in an out-of-distribution evaluation on the Music Arena dataset, comprising pairwise human preference battles across eleven generative music systems unseen during training, including aesthetic features consistently improves preference prediction, demonstrating strong generalisation of the learned representations across generative architectures.
Abstract (translated)
URL
https://arxiv.org/abs/2605.03395
PDF
-
DECKER: Domain-invariant Embedding for Cross-Keyboard Extraction and Recognition
2026-05-05 05:43:24Bikrant Bikram Pratap Maurya, Nitin Choudhury, Daksh Agarwal, Arun Balaji BuduruAbstract
Acoustic side-channel attacks (ASCA) on keyboards pose a significant security risk, as keystrokes can be inferred from typing acoustics, revealing sensitive information. Prior ASCA studies are limited by small-scale datasets with restricted diversity in users, keyboards, and environments, constraining analysis across devices, microphones, and noise conditions. We introduce HEAR, a dataset designed to study ASCA along three axes: keyboard generalization, noise adaptation, and user bias. HEAR contains recordings from 53 participants using 37 laptop keyboards, collected in three realistic settings: (1) external microphone capture, (2) device microphone capture without network noise, and (3) VoIP-based streaming capture. This enables controlled evaluation across users, keyboards, and environments. On HEAR, we establish an ASCA benchmark spanning conventional features and pre-trained representations from raw audio and spectrograms in unimodal and multimodal settings. We propose DECKER, a domain-invariant keystroke inference framework with four stages: (1) Keyboard Signature Normalization to reduce device coloration, (2) domain-adversarial disentanglement to suppress keyboard identity, (3) supervised cross-keyboard contrastive alignment to enforce key consistency, and (4) Acoustic Style Randomization to synthesize unseen keyboard responses. We further explore sentence-level inference using an LLM-based post-processing layer to refine keystroke sequences via linguistic context. Results on HEAR show DECKER improves keystroke identification over strong baselines, particularly in cross-keyboard and cross-user settings, with further gains from language-model rectification. These findings highlight that ASCA remains effective across diverse users, devices, and noisy environments, underscoring its practical security risk.
Abstract (translated)
URL
https://arxiv.org/abs/2605.03384
PDF
-
Contrastive Regularization for Accent-Robust ASR
2026-05-05 02:35:44Van-Phat Thai, Aradhya Dhruv, Duc-Thinh Pham, Sameer AlamAbstract
ASR systems based on self-supervised acoustic pretraining and CTC fine-tuning achieve strong performance on native speech but remain sensitive to accent variability. We investigate supervised contrastive learning (SupCon) as a lightweight, accent-invariant auxiliary objective for CTC fine-tuning. An utterance-level contrastive loss regularizes encoder representations without architectural modification or explicit accent supervision. Experiments on the L2-ARCTIC benchmark show consistent WER reductions across multiple pretrained encoders, with up to 25 -- 29\% relative reduction under unseen-accent evaluation. Analysis using within-transcript cosine dispersion indicates that SupCon promotes more compact and stable representation geometry under accent variability. Overall, SupCon provides an effective and model-agnostic regularization strategy for improving accent robustness.
Abstract (translated)
URL
https://arxiv.org/abs/2605.03297
PDF
-
Phoneme-Level Deepfake Detection Across Emotional Conditions Using Self-Supervised Embeddings
2026-05-04 18:49:29Vamshi Nallaguntla, Shruti Kshirsagar, Anderson R. AvilaAbstract
Recent advances in emotional voice conversion (EVC) have enabled the generation of expressive synthetic speech, raising new concerns in audio deepfake detection. Existing approaches treat speech as a homogeneous signal and largely overlook its internal phonetic structure, limiting their interpretability in emotionally conditioned settings. In this work, we propose a phoneme-level framework to analyze emotionally manipulated synthetic speech using real and EVC-generated speech under matched emotional conditions with shared transcripts, phoneme-aligned TextGrids, and WavLM-based embeddings. Our results show that phoneme behavior varies across categories, with complex vowels and fricatives exhibiting higher divergence while simpler phonemes remain more stable. Phonemes with larger distributional differences are also found to be more easily detected, consistently across multiple emotions and synthesis systems. These findings demonstrate that phoneme-level analysis is an effective and interpretable approach for detecting emotionally manipulated synthetic speech.
Abstract (translated)
URL
https://arxiv.org/abs/2605.03079
PDF
-
Mixed-Precision Information Bottlenecks for On-Device Trait-State Disentanglement in Bipolar Agitation Detection
Abstract
Continuous monitoring of bipolar disorder agitation via voice biomarkers requires disentangling stable speaker traits from volatile affective states on resource-constrained edge devices. We introduce MP-IB, the first framework to treat mixed-precision quantization as an information bottleneck for clinical trait-state separation. The core insight is that numerical precision itself controls capacity: an FP16 trait head (1,024 bits) encodes speaker identity, while an INT4 state head (128 bits) captures agitation, yielding 8x information asymmetry without adversarial training. We augment this with Dynamic Precision Scheduling and Multi-Scale Temporal Fusion. On Bridge2AI-Voice (N=833, 4 sessions/participant, strict speaker-independent CV), MP-IB achieves rho = 0.117 (95\% CI: [0.089, 0.145], p=0.003 vs. chance), outperforming 94M-parameter WavLM-Adapter with in-domain SSL continuation (rho = -0.042), beta VAE disentanglement (rho = 0.089), and hand-crafted prosody (rho = 0.031) by 2.8--15.9 points absolute. Zero-shot transfer to CREMA-D achieves AUC=0.817. Identity leakage is suppressed to near-random (EER=0.42, MIA-AUC=0.52). End-to-end latency is 23.4 ms with a 617 KB footprint, enabling real-time monitoring on sub 20 dollar devices.
Abstract (translated)
URL
https://arxiv.org/abs/2605.03039
PDF
-
Liberating LLM Capabilities in Full-Duplex Speech Models
2026-05-04 17:54:41Luoyuan Zhang, Bokai Xu, Junbo Cui, Weiyue Sun, Yingjing Xu, Hanyu Liu, Yuan YaoAbstract
Speech-based large language models are typically constrained to spoken replies, which limits their user-facing outputs to what can be verbalized and suppresses text-native capabilities such as code generation, structured analysis, and multi-step reasoning in realtime interaction, for tasks that require persistent, structured, and inspectable intermediate outputs. Existing work improves spoken reasoning or full-duplex turn-taking, but still treats text as a hidden intermediate state or a subordinate modality rather than a first-class output channel. We propose Listen-Write-Speak (LWS), a text-first tri-channel paradigm in which a single autoregressive LLM continuously listens to user audio, writes visible free-form text as its primary output, and speaks a realtime oral response in parallel under a shared causal attention context. This behavior is implemented entirely through a Token Schema, requiring no architectural modifications, and learned via a two-stage data pipeline that synthesizes per-second cognitive annotations consistent with the revealed input timeline. Empirically, LWS demonstrates strong full-duplex interaction on Full-Duplex-Bench, reaches 4.72 on VoiceBench AlpacaEval, achieves 92.6% writing-speaking consistency, and consistently outperforms its internal ablations on URO-Bench. These results suggest that visible writing can serve as a first-class output channel for speech interaction without sacrificing realtime responsiveness. The code and dataset are available on the project page: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2606.07547
PDF
-
Private Speech Classification without Collapse: Stabilized DP Training and Offline Distillation
2026-05-04 15:19:02Yadi Wen, Tianxin Li, Enji Liang, Rong Du, Yue FuAbstract
We study example-level private supervised speech classification under a practical release constraint: training may access privileged side information, but the released model must be audio-only. This setting is important because speech systems can often exploit richer side information during development, whereas deployment and release require a lightweight unimodal model with auditable privacy guarantees. Using DP-SGD on the private dataset $D_{\text{priv}}$, we identify a strong-privacy failure mode ($\epsilon \le 1$) on imbalanced tasks, where training may collapse to a near single-class predictor, a phenomenon that overall accuracy can obscure. We therefore emphasize Macro-F1, balanced accuracy, and a simple collapse diagnostic. This failure is especially problematic in our release setting because a collapsed private teacher cannot provide useful supervision for the downstream audio-only student. To address this setting under strong privacy, we propose a two-stage protocol: (i) train a (possibly multimodal) DP teacher on $D_{\text{priv}}$, and (ii) distill an audio-only student on a fixed, recording-disjoint auxiliary dataset $D_{\text{aux}}$ using one-shot offline teacher probability outputs, releasing only the student. The DP guarantee applies only to $D_{\text{priv}}$; we make no DP claim for $D_{\text{aux}}$, and privacy of the released student with respect to $D_{\text{priv}}$ follows by post-processing. We frame this setting as involving four coupled bottlenecks: speech-induced optimization instability under DP-SGD, minority-class erosion under clipping and noise, teacher over-reliance on privileged modalities unavailable at deployment, and train--deploy modality mismatch. We address them with a DP-stabilizing acoustic front-end (DSAF), minibatch-adaptive bounded loss reweighting (AW-DP), privileged-modality dropout, and offline teacher-to-student distillation.
Abstract (translated)
URL
https://arxiv.org/abs/2605.02718
PDF
-
RenCon 2025: Revival of the Expressive Performance Rendering Competition
2026-05-03 21:06:08Huan Zhang, Taegyun Kwon, Anders Friburg, Junyan Jiang, Hayeon Bang, Hyeyoon Cho, Gus Xia, Akira Maezawa, Simon Dixon, Dasaem JeongAbstract
This paper presents a comprehensive documentation of RenCon 2025, the revival of the expressive performance rendering competition which took place at ISMIR 2025 in Daejeon, Korea. The competition attracted 9 entries from international research groups, representing diverse approaches to expressive piano performance rendering. The two-phase assessment structure comprised a preliminary online evaluation and live real-time rendering at the conference. We analyze the competition format, participant demographics, system performance, and lessons learned for future iterations. The results demonstrate significant advances in expressive rendering capabilities while highlighting remaining challenges in achieving human-level musical expression.
Abstract (translated)
URL
https://arxiv.org/abs/2605.02059
PDF
-
TMD-Bench: A Multi-Level Evaluation Paradigm for Music-Dance Co-Generation
2026-05-03 10:25:47Xiaoda Yang, Majun Zhang, Changhao Pan, Nick Huang, Yang Yuguang, Fan Zhuo, Pengfei Zhou, Jin Zhou, Sizhe Shan, Shan Yang, Miles Yang, Yang You, Zhou ZhaoAbstract
Unified audio-visual generation is rapidly gaining industrial and creative relevance, enabling applications in virtual production and interactive media. However, when moving from general audio-video synthesis to music-dance co-generation, the task becomes substantially harder: musical rhythm, phrasing, and accents must drive choreographic motion at fine temporal resolution, and such rhythmic coupling is not captured by unimodal metrics or generic audiovisual consistency scores used in current evaluation practice. We introduce TMD-Bench, a benchmark for text-driven music-dance co-generation that assesses systems across unimodal generation quality, instruction adherence, and cross-modal rhythmic alignment. The benchmark integrates computable physical metrics with perceptual multimodal judgments, and is supported by a curated rhythm-aligned music-dance dataset and a fine-grained Music Captioner for structured music semantics. TMD-Bench further reveals that (i) modern commercial audio-visual models, such as Veo 3 and Sora 2, produce high-quality music and video, while rhythmic coupling remains less consistently optimized and leaves room for improvement, and (ii) our unified baseline RhyJAM trained on rhythm-aligned data achieves competitive beat-level synchronization while maintaining competitive unimodal fidelity. This presents prospects for building next-generation music-dance models that explicitly optimize rhythmic and kinetic coherence.
Abstract (translated)
URL
https://arxiv.org/abs/2605.01809
PDF
-
Khala: Scaling Acoustic Token Language Models Toward High-Fidelity Music Generation
2026-05-03 09:13:20Jiafeng Liu, Yuanliang Dong, Hongjia Liu, Yuqing Cheng, Zhancheng Guo, Huijing Liang, Wenbo Zhan, Yuming Sun, Xiaobing Li, Feng Yu, Maosong SunAbstract
A common design pattern in high-quality music generation is to handle structure and fidelity in different representation spaces: a generator first models high-level structure, followed by diffusion-based or neural decoding stages that reconstruct fine details. In this work, we explore an alternative view: both may be progressively modeled within a single deep acoustic-token hierarchy. To study this, we build a 64-layer residual vector quantization (RVQ) acoustic representation and propose a two-stage coarse-to-fine generation framework. A backbone model first generates coarse acoustic tokens for the full track, and a super-resolution model then completes finer tokens within the same acoustic token space. The super-resolution stage works at full-track scale and refines tokens layer by layer while running in parallel over time, leading to a fixed 62-step inference process. To jointly improve lyric alignment and fine-detail reconstruction, we further introduce hybrid-attention training: the alignment objective uses causal attention, while layer-wise refinement uses full attention. A key finding is that text--vocal alignment can emerge within pure acoustic-token language modeling, without requiring a separate semantic token stage. Moreover, initializing the super-resolution model from the trained backbone significantly improves convergence and final quality. Taken together, our results suggest that high-quality music generation can be effectively pursued without separating structure and fidelity into heterogeneous representation spaces. Instead, both can be progressively modeled within a unified acoustic-token hierarchy, pointing toward a simpler and more unified path to high-quality music generation.
Abstract (translated)
URL
https://arxiv.org/abs/2605.01790
PDF
