Shan Yang

Jakob Wasserthal, Hanns-Christian Breit, Manfred T. Meyer et al.

We present a deep learning segmentation model that can automatically and robustly segment all major anatomical structures in body CT images. In this retrospective study, 1204 CT examinations (from the years 2012, 2016, and 2020) were used to segment 104 anatomical structures (27 organs, 59 bones, 10 muscles, 8 vessels) relevant for use cases such as organ volumetry, disease characterization, and surgical or radiotherapy planning. The CT images were randomly sampled from routine clinical studies and thus represent a real-world dataset (different ages, pathologies, scanners, body parts, sequences, and sites). The authors trained an nnU-Net segmentation algorithm on this dataset and calculated Dice similarity coefficients (Dice) to evaluate the model's performance. The trained algorithm was applied to a second dataset of 4004 whole-body CT examinations to investigate age dependent volume and attenuation changes. The proposed model showed a high Dice score (0.943) on the test set, which included a wide range of clinical data with major pathologies. The model significantly outperformed another publicly available segmentation model on a separate dataset (Dice score, 0.932 versus 0.871, respectively). The aging study demonstrated significant correlations between age and volume and mean attenuation for a variety of organ groups (e.g., age and aortic volume; age and mean attenuation of the autochthonous dorsal musculature). The developed model enables robust and accurate segmentation of 104 anatomical structures. The annotated dataset (https://doi.org/10.5281/zenodo.6802613) and toolkit (https://www.github.com/wasserth/TotalSegmentator) are publicly available.

Junyi Qiu, Lei Li, Sihan Wang et al.

Myocardial pathology segmentation (MyoPS) can be a prerequisite for the accurate diagnosis and treatment planning of myocardial infarction. However, achieving this segmentation is challenging, mainly due to the inadequate and indistinct information from an image. In this work, we develop an end-to-end deep neural network, referred to as MyoPS-Net, to flexibly combine five-sequence cardiac magnetic resonance (CMR) images for MyoPS. To extract precise and adequate information, we design an effective yet flexible architecture to extract and fuse cross-modal features. This architecture can tackle different numbers of CMR images and complex combinations of modalities, with output branches targeting specific pathologies. To impose anatomical knowledge on the segmentation results, we first propose a module to regularize myocardium consistency and localize the pathologies, and then introduce an inclusiveness loss to utilize relations between myocardial scars and edema. We evaluated the proposed MyoPS-Net on two datasets, i.e., a private one consisting of 50 paired multi-sequence CMR images and a public one from MICCAI2020 MyoPS Challenge. Experimental results showed that MyoPS-Net could achieve state-of-the-art performance in various scenarios. Note that in practical clinics, the subjects may not have full sequences, such as missing LGE CMR or mapping CMR scans. We therefore conducted extensive experiments to investigate the performance of the proposed method in dealing with such complex combinations of different CMR sequences. Results proved the superiority and generalizability of MyoPS-Net, and more importantly, indicated a practical clinical application.

Haitao Li, Tian Tan, Yuguang Yang et al.

The rapid advancement of instruction-guided audio generation has highlighted the critical need for robust alignment evaluation. Current automated evaluation methods heavily rely on holistic scoring from general-purpose large language models, which struggle to decouple complex instructions, lack interpretability, and fail to capture fine-grained attribute mismatches. To address this, we introduce a novel dynamic rubric-based evaluation paradigm that adaptively decomposes complex audio captions into a variable number of independent, verifiable binary rubric items. To rigorously benchmark this capability, we propose the AnyAudio-Judge Bench, a comprehensive, bilingual benchmark comprising 7,920 meticulously curated samples across four diverse audio domains (speech, sound, music, and mixed), featuring deliberately constructed hard negatives. Furthermore, we construct a large-scale corpus of 105K samples with explicit Chain-of-Thought (CoT) rationales to train our dedicated evaluator, the AnyAudio-Judge model. By employing a training pipeline that combines Supervised Fine-Tuning (SFT) and Group Relative Policy Optimization (GRPO), our model successfully aligns its reasoning paths with the rubric-based scoring mechanism. Extensive experiments demonstrate that AnyAudio-Judge not only significantly enhances zero-shot alignment detection compared to state-of-the-art baselines, but also provides precise and interpretable reward signals that substantially improve instruction alignment in downstream reinforcement learning for audio generation.

Wangbin Ding, Lei Li, Junyi Qiu et al.

Myocardial pathology segmentation (MyoPS) is critical for the risk stratification and treatment planning of myocardial infarction (MI). Multi-sequence cardiac magnetic resonance (MS-CMR) images can provide valuable information. For instance, balanced steady-state free precession cine sequences present clear anatomical boundaries, while late gadolinium enhancement and T2-weighted CMR sequences visualize myocardial scar and edema of MI, respectively. Existing methods usually fuse anatomical and pathological information from different CMR sequences for MyoPS, but assume that these images have been spatially aligned. However, MS-CMR images are usually unaligned due to the respiratory motions in clinical practices, which poses additional challenges for MyoPS. This work presents an automatic MyoPS framework for unaligned MS-CMR images. Specifically, we design a combined computing model for simultaneous image registration and information fusion, which aggregates multi-sequence features into a common space to extract anatomical structures (i.e., myocardium). Consequently, we can highlight the informative regions in the common space via the extracted myocardium to improve MyoPS performance, considering the spatial relationship between myocardial pathologies and myocardium. Experiments on a private MS-CMR dataset and a public dataset from the MYOPS2020 challenge show that our framework could achieve promising performance for fully automatic MyoPS.

Akshay Gadi Patil, Yiming Qian, Shan Yang et al.

The dominant majority of 3D models that appear in gaming, VR/AR, and those we use to train geometric deep learning algorithms are incomplete, since they are modeled as surface meshes and missing their interior structures. We present a learning framework to recover the shape interiors (RoSI) of existing 3D models with only their exteriors from multi-view and multi-articulation images. Given a set of RGB images that capture a target 3D object in different articulated poses, possibly from only few views, our method infers the interior planes that are observable in the input images. Our neural architecture is trained in a category-agnostic manner and it consists of a motion-aware multi-view analysis phase including pose, depth, and motion estimations, followed by interior plane detection in images and 3D space, and finally multi-view plane fusion. In addition, our method also predicts part articulations and is able to realize and even extrapolate the captured motions on the target 3D object. We evaluate our method by quantitative and qualitative comparisons to baselines and alternative solutions, as well as testing on untrained object categories and real image inputs to assess its generalization capabilities.

Ashraya K. Indrakanti, Jakob Wasserthal, Martin Segeroth et al.

Purpose: To develop an open-source nnU-Net-based AI model for combined detection and segmentation of unruptured intracranial aneurysms (UICA) in 3D TOF-MRI, and compare models trained on datasets with aneurysm-like differential diagnoses. Methods: This retrospective study (2020-2023) included 385 anonymized 3D TOF-MRI images from 364 patients (mean age 59 years, 60% female) at multiple centers plus 113 subjects from the ADAM challenge. Images featured untreated or possible UICAs and differential diagnoses. Four distinct training datasets were created, and the nnU-Net framework was used for model development. Performance was assessed on a separate test set using sensitivity and False Positive (FP)/case rate for detection, and DICE score and NSD (Normalized Surface Distance) with a 0.5mm threshold for segmentation. Statistical analysis included chi-square, Mann-Whitney-U, and Kruskal-Wallis tests, with significance set at p < 0.05. Results: Models achieved overall sensitivity between 82% and 85% and a FP/case rate of 0.20 to 0.31, with no significant differences (p = 0.90 and p = 0.16). The primary model showed 85% sensitivity and 0.23 FP/case rate, outperforming the ADAM-challenge winner (61%) and a nnU-Net trained on ADAM data (51%) in sensitivity (p < 0.05). It achieved a mean DICE score of 0.73 and an NSD of 0.84 for correctly detected UICA. Conclusions: Our open-source, nnU-Net-based AI model (available at 10.5281/zenodo.13386859) demonstrates high sensitivity, low false positive rates, and consistent segmentation accuracy for UICA detection and segmentation in 3D TOF-MRI, suggesting its potential to improve clinical diagnosis and for monitoring of UICA.

Jiayang Bai, Zhen He, Shan Yang et al.

Predicting panoramic indoor lighting from a single perspective image is a fundamental but highly ill-posed problem in computer vision and graphics. To achieve locale-aware and robust prediction, this problem can be decomposed into three sub-tasks: depth-based image warping, panorama inpainting and high-dynamic-range (HDR) reconstruction, among which the success of panorama inpainting plays a key role. Recent methods mostly rely on convolutional neural networks (CNNs) to fill the missing contents in the warped panorama. However, they usually achieve suboptimal performance since the missing contents occupy a very large portion in the panoramic space while CNNs are plagued by limited receptive fields. The spatially-varying distortion in the spherical signals further increases the difficulty for conventional CNNs. To address these issues, we propose a local-to-global strategy for large-scale panorama inpainting. In our method, a depth-guided local inpainting is first applied on the warped panorama to fill small but dense holes. Then, a transformer-based network, dubbed PanoTransformer, is designed to hallucinate reasonable global structures in the large holes. To avoid distortion, we further employ cubemap projection in our design of PanoTransformer. The high-quality panorama recovered at any locale helps us to capture spatially-varying indoor illumination with physically-plausible global structures and fine details.

Guanrou Yang, Tian Tan, Qian Chen et al.

Integrating speech understanding and generation is a pivotal step toward building unified speech models. However, the different representations required for these two tasks currently pose significant compatibility challenges. Typically, semantics-oriented features are learned from self-supervised learning (SSL), and acoustic-oriented features from reconstruction. Such fragmented representations hinder the realization of truly unified speech systems. We present WavCube, a compact continuous latent derived from an SSL speech encoder that simultaneously supports speech understanding, reconstruction, and generation. WavCube employs a two-stage training scheme. Stage 1 trains a semantic bottleneck to filter off-manifold redundancy that makes raw SSL features intractable for diffusion. Stage 2 injects fine-grained acoustic details via end-to-end reconstruction, while a semantic anchoring loss ensures the representation remains grounded within its original semantic manifold. Comprehensive experiments show that WavCube closely approaches WavLM performance on SUPERB despite an 8x dimensional compression, attains reconstruction quality on par with existing acoustic representations, delivers state-of-the-art zero-shot TTS performance with markedly faster training convergence, and excels in speech enhancement, separation, and voice conversion tasks on the SUPERB-SG benchmark. Systematic ablations reveal that WavCube's two-stage recipe resolves two intrinsic flaws of SSL features for generative modeling, paving the way for future unified speech systems. Codes and checkpoints are available at https://github.com/yanghaha0908/WavCube.

Wenfu Wang, Chenxing Li, Liqiang Zhang et al.

In this work, we present Covo-Audio, a 7B-parameter end-to-end LALM that directly processes continuous audio inputs and generates audio outputs within a single unified architecture. Through large-scale curated pretraining and targeted post-training, Covo-Audio achieves state-of-the-art or competitive performance among models of comparable scale across a broad spectrum of tasks, including speech-text modeling, spoken dialogue, speech understanding, audio understanding, and full-duplex voice interaction. Extensive evaluations demonstrate that the pretrained foundation model exhibits strong speech-text comprehension and semantic reasoning capabilities on multiple benchmarks, outperforming representative open-source models of comparable scale. Furthermore, Covo-Audio-Chat, the dialogue-oriented variant, demonstrates strong spoken conversational abilities, including understanding, contextual reasoning, instruction following, and generating contextually appropriate and empathetic responses, validating its applicability to real-world conversational assistant scenarios. Covo-Audio-Chat-FD, the evolved full-duplex model, achieves substantially superior performance on both spoken dialogue capabilities and full-duplex interaction behaviors, demonstrating its competence in practical robustness. To mitigate the high cost of deploying end-to-end LALMs for natural conversational systems, we propose an intelligence-speaker decoupling strategy that separates dialogue intelligence from voice rendering, enabling flexible voice customization with minimal text-to-speech (TTS) data while preserving dialogue performance. Overall, our results highlight the strong potential of 7B-scale models to integrate sophisticated audio intelligence with high-level semantic reasoning, and suggest a scalable path toward more capable and versatile LALMs.

Peihao Wang, Shan Yang, Xijun Wang et al.

Associative memory has long underpinned the design of sequential models. Beyond recall, humans reason by projecting future states and selecting goal-directed actions, a capability that modern language models increasingly require but do not natively encode. While prior work uses reinforcement learning or test-time training, planning remains external to the model architecture. We formulate reasoning as optimal control and introduce the Test-Time Control (TTC) layer, which performs finite-horizon LQR planning over latent states at inference time, represents a value function within neural architectures, and leverages it as the nested objective to enable planning before prediction. To ensure scalability, we derive a hardware-efficient LQR solver based on a symplectic formulation and implement it as a fused CUDA kernel, enabling parallel execution with minimal overhead. Integrated as an adapter into pretrained LLMs, TTC layers improve mathematical reasoning performance by up to +27.8% on MATH-500 and 2-3x Pass@8 improvements on AMC and AIME, demonstrating that embedding optimal control as an architectural component provides an effective and scalable mechanism for reasoning beyond test-time training.

Xijun Wang, Junbang Liang, Chun-Kai Wang et al.

In this work, we propose an efficient Video-Language Alignment (ViLA) network. Our ViLA model addresses both efficient frame sampling and effective cross-modal alignment in a unified way. In our ViLA network, we design a new learnable text-guided Frame-Prompter together with a new cross-modal distillation (QFormer-Distiller) module. Pre-trained large image-language models have shown promising results on problems such as visual question answering (VQA). However, how to efficiently and effectively sample video frames when adapting pre-trained large image-language model to video-language alignment is still the major challenge. Compared with prior work, our ViLA model demonstrates the capability of selecting key frames with critical contents, thus improving the video-language alignment accuracy while reducing the inference latency +3.3% on NExT-QA Temporal with 3.0X speed up). Overall, our ViLA network outperforms the state-of-the-art methods on the video question-answering benchmarks: +4.6% on STAR Interaction, +2.2% on STAR average with 3.0X speed up, ours 2-frames out-perform SeViLA 4-frames on the VLEP dataset with 4.2X speed-up. The code will be available at https://github.com/xijun-cs/ViLA.

Xijun Wang, Anqi Liang, Junbang Liang et al.

Scene-aware Complementary Item Retrieval (CIR) is a challenging task which requires to generate a set of compatible items across domains. Due to the subjectivity, it is difficult to set up a rigorous standard for both data collection and learning objectives. To address this challenging task, we propose a visual compatibility concept, composed of similarity (resembling in color, geometry, texture, and etc.) and complementarity (different items like table vs chair completing a group). Based on this notion, we propose a compatibility learning framework, a category-aware Flexible Bidirectional Transformer (FBT), for visual "scene-based set compatibility reasoning" with the cross-domain visual similarity input and auto-regressive complementary item generation. We introduce a "Flexible Bidirectional Transformer (FBT)" consisting of an encoder with flexible masking, a category prediction arm, and an auto-regressive visual embedding prediction arm. And the inputs for FBT are cross-domain visual similarity invariant embeddings, making this framework quite generalizable. Furthermore, our proposed FBT model learns the inter-object compatibility from a large set of scene images in a self-supervised way. Compared with the SOTA methods, this approach achieves up to 5.3% and 9.6% in FITB score and 22.3% and 31.8% SFID improvement on fashion and furniture, respectively.

Muhammad Osama Khan, Junbang Liang, Chun-Kai Wang et al.

Pre-training has been an important ingredient in developing strong monocular depth estimation models in recent years. For instance, self-supervised learning (SSL) is particularly effective by alleviating the need for large datasets with dense ground-truth depth maps. However, despite these improvements, our study reveals that the later layers of the SOTA SSL method are actually suboptimal. By examining the layer-wise representations, we demonstrate significant changes in these later layers during fine-tuning, indicating the ineffectiveness of their pre-trained features for depth estimation. To address these limitations, we propose MeSa, a comprehensive framework that leverages the complementary strengths of masked, geometric, and supervised pre-training. Hence, MeSa benefits from not only general-purpose representations learnt via masked pre training but also specialized depth-specific features acquired via geometric and supervised pre-training. Our CKA layer-wise analysis confirms that our pre-training strategy indeed produces improved representations for the later layers, overcoming the drawbacks of the SOTA SSL method. Furthermore, via experiments on the NYUv2 and IBims-1 datasets, we demonstrate that these enhanced representations translate to performance improvements in both the in-distribution and out-of-distribution settings. We also investigate the influence of the pre-training dataset and demonstrate the efficacy of pre-training on LSUN, which yields significantly better pre-trained representations. Overall, our approach surpasses the masked pre-training SSL method by a substantial margin of 17.1% on the RMSE. Moreover, even without utilizing any recently proposed techniques, MeSa also outperforms the most recent methods and establishes a new state-of-the-art for monocular depth estimation on the challenging NYUv2 dataset.

Shan Yang

We audit the multimodal-physics evaluation pipeline end-to-end and document three undetected construction practices that distort how the field measures vision-language reasoning: train-eval contamination, translation drift, and MCQ saturation. (1) Public training pools (UGPhysics-Train, SciInstruct, MMK12) pass single-stage 5-gram-Jaccard audits with zero hits across all six public physics evals; a three-stage audit (Jaccard -> mxbai-embed-large cosine -> Haiku-4.5 LLM-judge) surfaces 134 near-duplicates and 4,846 paraphrase candidates in SciInstruct alone. (2) A 17-pp Sonnet 4.5 delta on 59 paired Estonian-English olympiad problems (30.5% vs. 13.6%; sign test p=0.011, McNemar p=0.021, paired bootstrap 95% CI [+5.1, +28.9] pp). (3) A 46-pp format-and-novelty gradient on identical Sonnet weights between MCQ (79.7% on PhyX) and open-ended olympiad evaluation (33.4% on PhysOlym-A). We release four artifacts addressing these gaps: PhysCorp-A (6,432-record three-stage-audited multimodal corpus), PhysR1Corp (2,268-record closed-form RL pool), PhysOlym-A (500-problem, 99.8% novel-source held-out olympiad eval with native difficulty labels and an EN/ET bilingual subset), and Physics-R1, a reference GSPO+DAPO recipe cold-started from Qwen3-VL-8B-Thinking. Across 3 seeds, Physics-R1 lifts the audited corpus over the 8B base by +18.3 pp on PhysOlym-A liberal (8.0 -> 26.3 +/- 1.7; 7.1 pp behind Sonnet 4.5), +15.7 pp on PhysReason (23.9 -> 39.6 +/- 6.4; ahead of Qwen3-VL-32B and Gemini 2.5 Pro), +6.9 pp on OlympiadBench-Physics (46.2 +/- 1.5), and +4.1 pp on PhyX MCQ (77.8 +/- 0.3).

Tianxin Xie, Wentao Lei, Guanjie Huang et al.

Text-to-audio-video (T2AV) generation underpins a wide range of applications demanding realistic audio-visual content, including virtual reality, world modeling, gaming, and filmmaking. However, existing T2AV models remain incapable of generating physically plausible sounds, primarily due to their limited understanding of physical principles. To situate current research progress, we present PhyAVBench, a challenging audio physics-sensitivity benchmark designed to systematically evaluate the audio physics grounding capabilities of existing T2AV models. PhyAVBench comprises 1,000 groups of paired text prompts with controlled physical variables that implicitly induce sound variations, enabling a fine-grained assessment of models' sensitivity to changes in underlying acoustic conditions. We term this evaluation paradigm the Audio-Physics Sensitivity Test (APST). Unlike prior benchmarks that primarily focus on audio-video synchronization, PhyAVBench explicitly evaluates models' understanding of the physical mechanisms underlying sound generation, covering 6 major audio physics dimensions, 4 daily scenarios (music, sound effects, speech, and their mix), and 50 fine-grained test points, ranging from fundamental aspects such as sound diffraction to more complex phenomena, e.g., Helmholtz resonance. Each test point consists of multiple groups of paired prompts, where each prompt is grounded by at least 20 newly recorded or collected real-world videos, thereby minimizing the risk of data leakage during model pre-training. Both prompts and videos are iteratively refined through rigorous human-involved error correction and quality control to ensure high quality. We argue that only models with a genuine grasp of audio-related physical principles can generate physically consistent audio-visual content. We hope PhyAVBench will stimulate future progress in this critical yet largely unexplored domain.

Guanjie Huang, Danny H. K. Tsang, Shan Yang et al.

Cued Speech (CS) is a visual communication system that combines lip-reading with hand coding to facilitate communication for individuals with hearing impairments. Automatic CS Recognition (ACSR) aims to convert CS hand gestures and lip movements into text via AI-driven methods. Traditionally, the temporal asynchrony between hand and lip movements requires the design of complex modules to facilitate effective multimodal fusion. However, constrained by limited data availability, current methods demonstrate insufficient capacity for adequately training these fusion mechanisms, resulting in suboptimal performance. Recently, multi-agent systems have shown promising capabilities in handling complex tasks with limited data availability. To this end, we propose the first collaborative multi-agent system for ACSR, named Cued-Agent. It integrates four specialized sub-agents: a Multimodal Large Language Model-based Hand Recognition agent that employs keyframe screening and CS expert prompt strategies to decode hand movements, a pretrained Transformer-based Lip Recognition agent that extracts lip features from the input video, a Hand Prompt Decoding agent that dynamically integrates hand prompts with lip features during inference in a training-free manner, and a Self-Correction Phoneme-to-Word agent that enables post-process and end-to-end conversion from phoneme sequences to natural language sentences for the first time through semantic refinement. To support this study, we expand the existing Mandarin CS dataset by collecting data from eight hearing-impaired cuers, establishing a mixed dataset of fourteen subjects. Extensive experiments demonstrate that our Cued-Agent performs superbly in both normal and hearing-impaired scenarios compared with state-of-the-art methods. The implementation is available at https://github.com/DennisHgj/Cued-Agent.

Shan Yang

Text-to-image generation models have revolutionized content creation, but diffusion-based vision-language models still face challenges in precisely controlling the shape, appearance, and positional placement of objects in generated images using text guidance alone. Existing global image editing models rely on additional masks or images as guidance to achieve layout control, often requiring retraining of the model. While local object-editing models allow modifications to object shapes, they lack the capability to control object positions. To address these limitations, we propose the Mask-free Training-free Object-Level Layout Control Diffusion Model (MFTF), which provides precise control over object positions without requiring additional masks or images. The MFTF model supports both single-object and multi-object positional adjustments, such as translation and rotation, while enabling simultaneous layout control and object semantic editing. The MFTF model employs a parallel denoising process for both the source and target diffusion models. During this process, attention masks are dynamically generated from the cross-attention layers of the source diffusion model and applied to queries from the self-attention layers to isolate objects. These queries, generated in the source diffusion model, are then adjusted according to the layout control parameters and re-injected into the self-attention layers of the target diffusion model. This approach ensures accurate and precise positional control of objects. Project source code available at https://github.com/syang-genai/MFTF.

Shan Yang, Yongfei Zhang, Guanglin Niu et al.

Few-shot relation extraction (FSRE) is of great importance in long-tail distribution problem, especially in special domain with low-resource data. Most existing FSRE algorithms fail to accurately classify the relations merely based on the information of the sentences together with the recognized entity pairs, due to limited samples and lack of knowledge. To address this problem, in this paper, we proposed a novel entity CONCEPT-enhanced FEw-shot Relation Extraction scheme (ConceptFERE), which introduces the inherent concepts of entities to provide clues for relation prediction and boost the relations classification performance. Firstly, a concept-sentence attention module is developed to select the most appropriate concept from multiple concepts of each entity by calculating the semantic similarity between sentences and concepts. Secondly, a self-attention based fusion module is presented to bridge the gap of concept embedding and sentence embedding from different semantic spaces. Extensive experiments on the FSRE benchmark dataset FewRel have demonstrated the effectiveness and the superiority of the proposed ConceptFERE scheme as compared to the state-of-the-art baselines. Code is available at https://github.com/LittleGuoKe/ConceptFERE.

Shan Yang, Yang Liu

Scaling cooperative multi-agent reinforcement learning (MARL) is fundamentally limited by cross-agent noise: when agents share a common reward, the actions of all $N$ agents jointly determine each agent's learning signal, so cross-agent noise grows with $N$. In the policy gradient setting, per-agent gradient estimate variance scales as $Θ(N)$, yielding sample complexity $\mathcal{O}(N/ε)$. We observe that many domains -- cloud computing, transportation, power systems -- have differentiable analytical models that prescribe efficient system states. In this work, we propose Descent-Guided Policy Gradient (DG-PG), a framework that constructs noise-free per-agent guidance gradients from these analytical models, decoupling each agent's gradient from the actions of all others. We prove that DG-PG reduces gradient variance from $Θ(N)$ to $\mathcal{O}(1)$, preserves the equilibria of the cooperative game, and achieves agent-independent sample complexity $\mathcal{O}(1/ε)$. On a heterogeneous cloud scheduling task with up to 200 agents, DG-PG converges within 10 episodes at every tested scale -- from $N=5$ to $N=200$ -- directly confirming the predicted scale-invariant complexity, while MAPPO and IPPO fail to converge under identical architectures.

Xiaoda Yang, Majun Zhang, Changhao Pan et al.

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.

Qinjuan Wang, Shan Yang, Yongli Zhu

This paper introduces an LLM agent that automates power grid static analysis by converting natural language into MATPOWER scripts. The framework utilizes DeepSeek-OCR to build an enhanced vector database from MATPOWER manuals. To ensure reliability, it devises a three-tier error-correction system: a static pre-check, a dynamic feedback loop, and a semantic validator. Operating via the Model Context Protocol, the tool enables asynchronous execution and automatically debugging in MATLAB. Experimental results demonstrate that the system achieves a 82.38% accuracy regarding the code fidelity, effectively eliminating hallucinations even in complex analysis tasks.

Zhendong Xiao, Changhao Chen, Shan Yang et al.

Camera relocalization is pivotal in computer vision, with applications in AR, drones, robotics, and autonomous driving. It estimates 3D camera position and orientation (6-DoF) from images. Unlike traditional methods like SLAM, recent strides use deep learning for direct end-to-end pose estimation. We propose EffLoc, a novel efficient Vision Transformer for single-image camera relocalization. EffLoc's hierarchical layout, memory-bound self-attention, and feed-forward layers boost memory efficiency and inter-channel communication. Our introduced sequential group attention (SGA) module enhances computational efficiency by diversifying input features, reducing redundancy, and expanding model capacity. EffLoc excels in efficiency and accuracy, outperforming prior methods, such as AtLoc and MapNet. It thrives on large-scale outdoor car-driving scenario, ensuring simplicity, end-to-end trainability, and eliminating handcrafted loss functions.

Guang Yang, Jingkun Chen, Xicheng Sheng et al.

Late gadolinium enhancement MRI (LGE MRI) is the gold standard for the detection of myocardial scars for post myocardial infarction (MI). LGE MRI requires the injection of a contrast agent, which carries potential side effects and increases scanning time and patient discomfort. To address these issues, we propose a novel framework that combines cardiac motion observed in cine MRI with image texture information to segment the myocardium and scar tissue in the left ventricle. Cardiac motion tracking can be formulated as a full cardiac image cycle registration problem, which can be solved via deep neural networks. Experimental results prove that the proposed method can achieve scar segmentation based on non-contrasted cine images with comparable accuracy to LGE MRI. This demonstrates its potential as an alternative to contrast-enhanced techniques for scar detection.

Le Xu, Chenxing Li, Yong Ren et al.

Current vision-guided audio captioning systems frequently fail to address audiovisual misalignment in real-world scenarios, such as dubbed content or off-screen sounds. To bridge this critical gap, we present an entropy-aware gated fusion framework that dynamically modulates visual information flow through cross-modal uncertainty quantification. Our novel approach employs attention entropy analysis in cross-attention layers to automatically identify and suppress misleading visual cues during modal fusion. Complementing this architecture, we develop a batch-wise audiovisual shuffling technique that generates synthetic mismatched training pairs, greatly enhancing model resilience against alignment noise. Evaluations on the AudioCaps benchmark demonstrate our system's superior performance over existing baselines, especially in mismatched modality scenarios. Furthermore, our solution demonstrates an approximately 6x improvement in inference speed compared to the baseline.

Shan Yang

Mean-field reinforcement learning (MF-RL) scales multi-agent RL to large populations by reducing each agent's dependence on others to a single summary statistic -- the mean action. However, this reduction requires every agent to act at every time step; when some agents are idle, the mean action is simply undefined. Addressing asynchrony therefore requires a different summary statistic -- one that remains defined regardless of which agents act. The population distribution $μ\in Δ(\mathcal{O})$ -- the fraction of agents at each observation -- satisfies this requirement: its dimension is independent of $N$, and under exchangeability it fully determines each agent's reward and transition. Existing MF-RL theory, however, is built on the mean action and does not extend to $μ$. We therefore construct the Temporal Mean Field (TMF) framework around the population distribution $μ$ from scratch, covering the full spectrum from fully synchronous to purely sequential decision-making within a single theory. We prove existence and uniqueness of TMF equilibria, establish an $O(1/\sqrt{N})$ finite-population approximation bound that holds regardless of how many agents act per step, and prove convergence of a policy gradient algorithm (TMF-PG) to the unique equilibrium. Experiments on a resource selection game and a dynamic queueing game confirm that TMF-PG achieves near-identical performance whether one agent or all $N$ act per step, with approximation error decaying at the predicted $O(1/\sqrt{N})$ rate.

Tianxin Xie, Shan Yang, Chenxing Li et al.

Text-to-speech (TTS) has shown great progress in recent years. However, most existing TTS systems offer only coarse and rigid emotion control, typically via discrete emotion labels or a carefully crafted and detailed emotional text prompt, making fine-grained emotion manipulation either inaccessible or unstable. These models also require extensive, high-quality datasets for training. To address these limitations, we propose EmoSteer-TTS, a novel training-free approach, to achieve fine-grained speech emotion control (conversion, interpolation, erasure) by activation steering. We first empirically observe that modifying a subset of the internal activations within a flow matching-based TTS model can effectively alter the emotional tone of synthesized speech. Building on this insight, we then develop a training-free and efficient algorithm, including activation extraction, emotional token searching, and inference-time steering, which can be seamlessly integrated into a wide range of pretrained models (e.g., F5-TTS, CosyVoice2, and E2-TTS). In addition, to derive effective steering vectors, we construct a curated emotional speech dataset with diverse speakers. Extensive experiments demonstrate that EmoSteer-TTS enables fine-grained, interpretable, and continuous control over speech emotion, outperforming the state-of-the-art (SOTA). To the best of our knowledge, this is the first method that achieves training-free and continuous fine-grained emotion control in TTS. Demo samples are available at https://emosteer-tts-demo.pages.dev/.

Zhendong Xiao, Wu Wei, Shujie Ji et al.

Camera relocalization, a cornerstone capability of modern computer vision, accurately determines a camera's position and orientation (6-DoF) from images and is essential for applications in augmented reality (AR), mixed reality (MR), autonomous driving, delivery drones, and robotic navigation. Unlike traditional deep learning-based methods that regress camera pose from images in a single scene, which often lack generalization and robustness in diverse environments, we propose MVL-Loc, a novel end-to-end multi-scene 6-DoF camera relocalization framework. MVL-Loc leverages pretrained world knowledge from vision-language models (VLMs) and incorporates multimodal data to generalize across both indoor and outdoor settings. Furthermore, natural language is employed as a directive tool to guide the multi-scene learning process, facilitating semantic understanding of complex scenes and capturing spatial relationships among objects. Extensive experiments on the 7Scenes and Cambridge Landmarks datasets demonstrate MVL-Loc's robustness and state-of-the-art performance in real-world multi-scene camera relocalization, with improved accuracy in both positional and orientational estimates.

Jinting Wang, Shan Yang, Chenxing Li et al.

Cued Speech (CS) enhances lipreading via hand coding, offering visual phonemic cues that support precise speech perception for the hearing-impaired. The task of CS Video-to-Speech generation (CSV2S) aims to convert CS videos into intelligible speech signals. Most existing research focuses on CS Recognition (CSR), which transcribes video content into text. Consequently, a common solution for CSV2S is to integrate CSR with a text-to-speech (TTS) system. However, this pipeline relies on text as an intermediate medium, which may lead to error propagation and temporal misalignment between speech and CS video dynamics. In contrast, directly generating audio speech from CS video (direct CSV2S) often suffers from the inherent multimodal complexity and the limited availability of CS data. To address these challenges, we propose UniCUE, the first unified framework for CSV2S that directly generates speech from CS videos without relying on intermediate text. The core innovation of UniCUE lies in integrating an understanding task (CSR) that provides fine-grained CS visual-semantic cues to guide speech generation. Specifically, UniCUE incorporates a pose-aware visual processor, a semantic alignment pool that enables precise visual-semantic mapping, and a VisioPhonetic adapter to bridge the understanding and generation tasks within a unified architecture. To support this framework, we construct UniCUE-HI, a large-scale Mandarin CS dataset containing 11282 videos from 14 cuers, including both hearing-impaired and normal-hearing individuals. Extensive experiments on this dataset demonstrate that UniCUE achieves state-of-the-art performance across multiple evaluation metrics.

Shan Yang, Yongli Zhu

This paper presents a study on using different offline reinforcement learning algorithms for microgrid voltage regulation with solar power penetration. When environment interaction is unviable due to technical or safety reasons, the proposed approach can still obtain an applicable model through offline-style training on a previously collected dataset, lowering the negative impact of lacking online environment interactions. Experiment results on the IEEE 33-bus system demonstrate the feasibility and effectiveness of the proposed approach on different offline datasets, including the one with merely low-quality experience.

Shan Yang, Yongfei Zhang

Multimodal large language models (MLLM) have achieved satisfactory results in many tasks. However, their performance in the task of ReID (ReID) has not been explored to date. This paper will investigate how to adapt them for the task of ReID. An intuitive idea is to fine-tune MLLM with ReID image-text datasets, and then use their visual encoder as a backbone for ReID. However, there still exist two apparent issues: (1) Designing instructions for ReID, MLLMs may overfit specific instructions, and designing a variety of instructions will lead to higher costs. (2) When fine-tuning the visual encoder of a MLLM, it is not trained synchronously with the ReID task. As a result, the effectiveness of the visual encoder fine-tuning cannot be directly reflected in the performance of the ReID task. To address these problems, this paper proposes MLLMReID: Multimodal Large Language Model-based ReID. Firstly, we proposed Common Instruction, a simple approach that leverages the essence ability of LLMs to continue writing, avoiding complex and diverse instruction design. Secondly, we propose a multi-task learning-based synchronization module to ensure that the visual encoder of the MLLM is trained synchronously with the ReID task. The experimental results demonstrate the superiority of our method.

Disong Wang, Shan Yang, Dan Su et al.

Though significant progress has been made for speaker-dependent Video-to-Speech (VTS) synthesis, little attention is devoted to multi-speaker VTS that can map silent video to speech, while allowing flexible control of speaker identity, all in a single system. This paper proposes a novel multi-speaker VTS system based on cross-modal knowledge transfer from voice conversion (VC), where vector quantization with contrastive predictive coding (VQCPC) is used for the content encoder of VC to derive discrete phoneme-like acoustic units, which are transferred to a Lip-to-Index (Lip2Ind) network to infer the index sequence of acoustic units. The Lip2Ind network can then substitute the content encoder of VC to form a multi-speaker VTS system to convert silent video to acoustic units for reconstructing accurate spoken content. The VTS system also inherits the advantages of VC by using a speaker encoder to produce speaker representations to effectively control the speaker identity of generated speech. Extensive evaluations verify the effectiveness of proposed approach, which can be applied in both constrained vocabulary and open vocabulary conditions, achieving state-of-the-art performance in generating high-quality speech with high naturalness, intelligibility and speaker similarity. Our demo page is released here: https://wendison.github.io/VCVTS-demo/

Jiayang Bai, Jie Guo, Chenchen Wan et al.

Lighting prediction from a single image is becoming increasingly important in many vision and augmented reality (AR) applications in which shading and shadow consistency between virtual and real objects should be guaranteed. However, this is a notoriously ill-posed problem, especially for indoor scenarios, because of the complexity of indoor luminaires and the limited information involved in 2D images. In this paper, we propose a graph learning-based framework for indoor lighting estimation. At its core is a new lighting model (dubbed DSGLight) based on depth-augmented Spherical Gaussians (SG) and a Graph Convolutional Network (GCN) that infers the new lighting representation from a single LDR image of limited field-of-view. Our lighting model builds 128 evenly distributed SGs over the indoor panorama, where each SG encoding the lighting and the depth around that node. The proposed GCN then learns the mapping from the input image to DSGLight. Compared with existing lighting models, our DSGLight encodes both direct lighting and indirect environmental lighting more faithfully and compactly. It also makes network training and inference more stable. The estimated depth distribution enables temporally stable shading and shadows under spatially-varying lighting. Through thorough experiments, we show that our method obviously outperforms existing methods both qualitatively and quantitatively.

Yi Lei, Shan Yang, Xinsheng Wang et al.

Expressive synthetic speech is essential for many human-computer interaction and audio broadcast scenarios, and thus synthesizing expressive speech has attracted much attention in recent years. Previous methods performed the expressive speech synthesis either with explicit labels or with a fixed-length style embedding extracted from reference audio, both of which can only learn an average style and thus ignores the multi-scale nature of speech prosody. In this paper, we propose MsEmoTTS, a multi-scale emotional speech synthesis framework, to model the emotion from different levels. Specifically, the proposed method is a typical attention-based sequence-to-sequence model and with proposed three modules, including global-level emotion presenting module (GM), utterance-level emotion presenting module (UM), and local-level emotion presenting module (LM), to model the global emotion category, utterance-level emotion variation, and syllable-level emotion strength, respectively. In addition to modeling the emotion from different levels, the proposed method also allows us to synthesize emotional speech in different ways, i.e., transferring the emotion from reference audio, predicting the emotion from input text, and controlling the emotion strength manually. Extensive experiments conducted on a Chinese emotional speech corpus demonstrate that the proposed method outperforms the compared reference audio-based and text-based emotional speech synthesis methods on the emotion transfer speech synthesis and text-based emotion prediction speech synthesis respectively. Besides, the experiments also show that the proposed method can control the emotion expressions flexibly. Detailed analysis shows the effectiveness of each module and the good design of the proposed method.

Hai Su, Shan Yang, Shuqing Zhang et al.

Color image steganography based on deep learning is the art of hiding information in the color image. Among them, image hiding steganography(hiding image with image) has attracted much attention in recent years because of its great steganographic capacity. However, images generated by image hiding steganography may show some obvious color distortion or artificial texture traces. We propose a color image steganographic model based on frequency sub-band selection to solve the above problems. Firstly, we discuss the relationship between the characteristics of different color spaces/frequency sub-bands and the generated image quality. Then, we select the B channel of the RGB image as the embedding channel and the high-frequency sub-band as the embedding domain. DWT(discrete wavelet transformation) transforms B channel information and secret gray image into frequency domain information, and then the secret image is embedded and extracted in the frequency domain. Comprehensive experiments demonstrate that images generated by our model have better image quality, and the imperceptibility is significantly increased.

Songxiang Liu, Shan Yang, Dan Su et al.

Cross-speaker style transfer (CSST) in text-to-speech (TTS) synthesis aims at transferring a speaking style to the synthesised speech in a target speaker's voice. Most previous CSST approaches rely on expensive high-quality data carrying desired speaking style during training and require a reference utterance to obtain speaking style descriptors as conditioning on the generation of a new sentence. This work presents Referee, a robust reference-free CSST approach for expressive TTS, which fully leverages low-quality data to learn speaking styles from text. Referee is built by cascading a text-to-style (T2S) model with a style-to-wave (S2W) model. Phonetic PosteriorGram (PPG), phoneme-level pitch and energy contours are adopted as fine-grained speaking style descriptors, which are predicted from text using the T2S model. A novel pretrain-refinement method is adopted to learn a robust T2S model by only using readily accessible low-quality data. The S2W model is trained with high-quality target data, which is adopted to effectively aggregate style descriptors and generate high-fidelity speech in the target speaker's voice. Experimental results are presented, showing that Referee outperforms a global-style-token (GST)-based baseline approach in CSST.

Arsha Nagrani, Shan Yang, Anurag Arnab et al.

Humans perceive the world by concurrently processing and fusing high-dimensional inputs from multiple modalities such as vision and audio. Machine perception models, in stark contrast, are typically modality-specific and optimised for unimodal benchmarks, and hence late-stage fusion of final representations or predictions from each modality (`late-fusion') is still a dominant paradigm for multimodal video classification. Instead, we introduce a novel transformer based architecture that uses `fusion bottlenecks' for modality fusion at multiple layers. Compared to traditional pairwise self-attention, our model forces information between different modalities to pass through a small number of bottleneck latents, requiring the model to collate and condense the most relevant information in each modality and only share what is necessary. We find that such a strategy improves fusion performance, at the same time reducing computational cost. We conduct thorough ablation studies, and achieve state-of-the-art results on multiple audio-visual classification benchmarks including Audioset, Epic-Kitchens and VGGSound. All code and models will be released.

Jian Cong, Shan Yang, Lei Xie et al.

Current two-stage TTS framework typically integrates an acoustic model with a vocoder -- the acoustic model predicts a low resolution intermediate representation such as Mel-spectrum while the vocoder generates waveform from the intermediate representation. Although the intermediate representation is served as a bridge, there still exists critical mismatch between the acoustic model and the vocoder as they are commonly separately learned and work on different distributions of representation, leading to inevitable artifacts in the synthesized speech. In this work, different from using pre-designed intermediate representation in most previous studies, we propose to use VAE combining with GAN to learn a latent representation directly from speech and then utilize a flow-based acoustic model to model the distribution of the latent representation from text. In this way, the mismatch problem is migrated as the two stages work on the same distribution. Results demonstrate that the flow-based acoustic model can exactly model the distribution of our learned speech representation and the proposed TTS framework, namely Glow-WaveGAN, can produce high fidelity speech outperforming the state-of-the-art GAN-based model.

Jian Cong, Shan Yang, Na Hu et al.

In spoken conversations, spontaneous behaviors like filled pause and prolongations always happen. Conversational partner tends to align features of their speech with their interlocutor which is known as entrainment. To produce human-like conversations, we propose a unified controllable spontaneous conversational speech synthesis framework to model the above two phenomena. Specifically, we use explicit labels to represent two typical spontaneous behaviors filled-pause and prolongation in the acoustic model and develop a neural network based predictor to predict the occurrences of the two behaviors from text. We subsequently develop an algorithm based on the predictor to control the occurrence frequency of the behaviors, making the synthesized speech vary from less disfluent to more disfluent. To model the speech entrainment at acoustic level, we utilize a context acoustic encoder to extract a global style embedding from the previous speech conditioning on the synthesizing of current speech. Furthermore, since the current and previous utterances belong to the different speakers in a conversation, we add a domain adversarial training module to eliminate the speaker-related information in the acoustic encoder while maintaining the style-related information. Experiments show that our proposed approach can synthesize realistic conversations and control the occurrences of the spontaneous behaviors naturally.

Bo Hu, Bryan Seybold, Shan Yang et al.

We present a method to infer the 3D pose of mice, including the limbs and feet, from monocular videos. Many human clinical conditions and their corresponding animal models result in abnormal motion, and accurately measuring 3D motion at scale offers insights into health. The 3D poses improve classification of health-related attributes over 2D representations. The inferred poses are accurate enough to estimate stride length even when the feet are mostly occluded. This method could be applied as part of a continuous monitoring system to non-invasively measure animal health.

Ruilong Li, Shan Yang, David A. Ross et al.

We present AIST++, a new multi-modal dataset of 3D dance motion and music, along with FACT, a Full-Attention Cross-modal Transformer network for generating 3D dance motion conditioned on music. The proposed AIST++ dataset contains 5.2 hours of 3D dance motion in 1408 sequences, covering 10 dance genres with multi-view videos with known camera poses -- the largest dataset of this kind to our knowledge. We show that naively applying sequence models such as transformers to this dataset for the task of music conditioned 3D motion generation does not produce satisfactory 3D motion that is well correlated with the input music. We overcome these shortcomings by introducing key changes in its architecture design and supervision: FACT model involves a deep cross-modal transformer block with full-attention that is trained to predict $N$ future motions. We empirically show that these changes are key factors in generating long sequences of realistic dance motion that are well-attuned to the input music. We conduct extensive experiments on AIST++ with user studies, where our method outperforms recent state-of-the-art methods both qualitatively and quantitatively.

Haohan Guo, Heng Lu, Na Hu et al.

This paper describes an end-to-end adversarial singing voice conversion (EA-SVC) approach. It can directly generate arbitrary singing waveform by given phonetic posteriorgram (PPG) representing content, F0 representing pitch, and speaker embedding representing timbre, respectively. Proposed system is composed of three modules: generator $G$, the audio generation discriminator $D_{A}$, and the feature disentanglement discriminator $D_F$. The generator $G$ encodes the features in parallel and inversely transforms them into the target waveform. In order to make timbre conversion more stable and controllable, speaker embedding is further decomposed to the weighted sum of a group of trainable vectors representing different timbre clusters. Further, to realize more robust and accurate singing conversion, disentanglement discriminator $D_F$ is proposed to remove pitch and timbre related information that remains in the encoded PPG. Finally, a two-stage training is conducted to keep a stable and effective adversarial training process. Subjective evaluation results demonstrate the effectiveness of our proposed methods. Proposed system outperforms conventional cascade approach and the WaveNet based end-to-end approach in terms of both singing quality and singer similarity. Further objective analysis reveals that the model trained with the proposed two-stage training strategy can produce a smoother and sharper formant which leads to higher audio quality.

Tao Li, Shan Yang, Liumeng Xue et al.

Emotion embedding space learned from references is a straightforward approach for emotion transfer in encoder-decoder structured emotional text to speech (TTS) systems. However, the transferred emotion in the synthetic speech is not accurate and expressive enough with emotion category confusions. Moreover, it is hard to select an appropriate reference to deliver desired emotion strength. To solve these problems, we propose a novel approach based on Tacotron. First, we plug two emotion classifiers -- one after the reference encoder, one after the decoder output -- to enhance the emotion-discriminative ability of the emotion embedding and the predicted mel-spectrum. Second, we adopt style loss to measure the difference between the generated and reference mel-spectrum. The emotion strength in the synthetic speech can be controlled by adjusting the value of the emotion embedding as the emotion embedding can be viewed as the feature map of the mel-spectrum. Experiments on emotion transfer and strength control have shown that the synthetic speech of the proposed method is more accurate and expressive with less emotion category confusions and the control of emotion strength is more salient to listeners.

Zhichao Wang, Wenshuo Ge, Xiong Wang et al.

This paper proposes an interesting voice and accent joint conversion approach, which can convert an arbitrary source speaker's voice to a target speaker with non-native accent. This problem is challenging as each target speaker only has training data in native accent and we need to disentangle accent and speaker information in the conversion model training and re-combine them in the conversion stage. In our recognition-synthesis conversion framework, we manage to solve this problem by two proposed tricks. First, we use accent-dependent speech recognizers to obtain bottleneck features for different accented speakers. This aims to wipe out other factors beyond the linguistic information in the BN features for conversion model training. Second, we propose to use adversarial training to better disentangle the speaker and accent information in our encoder-decoder based conversion model. Specifically, we plug an auxiliary speaker classifier to the encoder, trained with an adversarial loss to wipe out speaker information from the encoder output. Experiments show that our approach is superior to the baseline. The proposed tricks are quite effective in improving accentedness and audio quality and speaker similarity are well maintained.

Yi Lei, Shan Yang, Lei Xie

This paper proposes a unified model to conduct emotion transfer, control and prediction for sequence-to-sequence based fine-grained emotional speech synthesis. Conventional emotional speech synthesis often needs manual labels or reference audio to determine the emotional expressions of synthesized speech. Such coarse labels cannot control the details of speech emotion, often resulting in an averaged emotion expression delivery, and it is also hard to choose suitable reference audio during inference. To conduct fine-grained emotion expression generation, we introduce phoneme-level emotion strength representations through a learned ranking function to describe the local emotion details, and the sentence-level emotion category is adopted to render the global emotions of synthesized speech. With the global render and local descriptors of emotions, we can obtain fine-grained emotion expressions from reference audio via its emotion descriptors (for transfer) or directly from phoneme-level manual labels (for control). As for the emotional speech synthesis with arbitrary text inputs, the proposed model can also predict phoneme-level emotion expressions from texts, which does not require any reference audio or manual label.

Heyang Xue, Shan Yang, Yi Lei et al.

Singing voice synthesis has been paid rising attention with the rapid development of speech synthesis area. In general, a studio-level singing corpus is usually necessary to produce a natural singing voice from lyrics and music-related transcription. However, such a corpus is difficult to collect since it's hard for many of us to sing like a professional singer. In this paper, we propose an approach -- Learn2Sing that only needs a singing teacher to generate the target speakers' singing voice without their singing voice data. In our approach, a teacher's singing corpus and speech from multiple target speakers are trained in a frame-level auto-regressive acoustic model where singing and speaking share the common speaker embedding and style tag embedding. Meanwhile, since there is no music-related transcription for the target speaker, we use log-scale fundamental frequency (LF0) as an auxiliary feature as the inputs of the acoustic model for building a unified input representation. In order to enable the target speaker to sing without singing reference audio in the inference stage, a duration model and an LF0 prediction model are also trained. Particularly, we employ domain adversarial training (DAT) in the acoustic model, which aims to enhance the singing performance of target speakers by disentangling style from acoustic features of singing and speaking data. Our experiments indicate that the proposed approach is capable of synthesizing singing voice for target speaker given only their speech samples.

Jian Cong, Shan Yang, Lei Xie et al.

Data efficient voice cloning aims at synthesizing target speaker's voice with only a few enrollment samples at hand. To this end, speaker adaptation and speaker encoding are two typical methods based on base model trained from multiple speakers. The former uses a small set of target speaker data to transfer the multi-speaker model to target speaker's voice through direct model update, while in the latter, only a few seconds of target speaker's audio directly goes through an extra speaker encoding model along with the multi-speaker model to synthesize target speaker's voice without model update. Nevertheless, the two methods need clean target speaker data. However, the samples provided by user may inevitably contain acoustic noise in real applications. It's still challenging to generating target voice with noisy data. In this paper, we study the data efficient voice cloning problem from noisy samples under the sequence-to-sequence based TTS paradigm. Specifically, we introduce domain adversarial training (DAT) to speaker adaptation and speaker encoding, which aims to disentangle noise from speech-noise mixture. Experiments show that for both speaker adaptation and encoding, the proposed approaches can consistently synthesize clean speech from noisy speaker samples, apparently outperforming the method adopting state-of-the-art speech enhancement module.

Fengyu Yang, Shan Yang, Qinghua Wu et al.

Attention-based seq2seq text-to-speech systems, especially those use self-attention networks (SAN), have achieved state-of-art performance. But an expressive corpus with rich prosody is still challenging to model as 1) prosodic aspects, which span across different sentential granularities and mainly determine acoustic expressiveness, are difficult to quantize and label and 2) the current seq2seq framework extracts prosodic information solely from a text encoder, which is easily collapsed to an averaged expression for expressive contents. In this paper, we propose a context extractor, which is built upon SAN-based text encoder, to sufficiently exploit the sentential context over an expressive corpus for seq2seq-based TTS. Our context extractor first collects prosodic-related sentential context information from different SAN layers and then aggregates them to learn a comprehensive sentence representation to enhance the expressiveness of the final generated speech. Specifically, we investigate two methods of context aggregation: 1) direct aggregation which directly concatenates the outputs of different SAN layers, and 2) weighted aggregation which uses multi-head attention to automatically learn contributions for different SAN layers. Experiments on two expressive corpora show that our approach can produce more natural speech with much richer prosodic variations, and weighted aggregation is more superior in modeling expressivity.

Geng Yang, Shan Yang, Kai Liu et al.

In this paper, we propose multi-band MelGAN, a much faster waveform generation model targeting to high-quality text-to-speech. Specifically, we improve the original MelGAN by the following aspects. First, we increase the receptive field of the generator, which is proven to be beneficial to speech generation. Second, we substitute the feature matching loss with the multi-resolution STFT loss to better measure the difference between fake and real speech. Together with pre-training, this improvement leads to both better quality and better training stability. More importantly, we extend MelGAN with multi-band processing: the generator takes mel-spectrograms as input and produces sub-band signals which are subsequently summed back to full-band signals as discriminator input. The proposed multi-band MelGAN has achieved high MOS of 4.34 and 4.22 in waveform generation and TTS, respectively. With only 1.91M parameters, our model effectively reduces the total computational complexity of the original MelGAN from 5.85 to 0.95 GFLOPS. Our Pytorch implementation, which will be open-resourced shortly, can achieve a real-time factor of 0.03 on CPU without hardware specific optimization.

Shan Yang, Yuxuan Wang, Lei Xie

Attention-based sequence-to-sequence (seq2seq) speech synthesis has achieved extraordinary performance. But a studio-quality corpus with manual transcription is necessary to train such seq2seq systems. In this paper, we propose an approach to build high-quality and stable seq2seq based speech synthesis system using challenging found data, where training speech contains noisy interferences (acoustic noise) and texts are imperfect speech recognition transcripts (textual noise). To deal with text-side noise, we propose a VQVAE based heuristic method to compensate erroneous linguistic feature with phonetic information learned directly from speech. As for the speech-side noise, we propose to learn a noise-independent feature in the auto-regressive decoder through adversarial training and data augmentation, which does not need an extra speech enhancement model. Experiments show the effectiveness of the proposed approach in dealing with text-side and speech-side noise. Surpassing the denoising approach based on a state-of-the-art speech enhancement model, our system built on noisy found data can synthesize clean and high-quality speech with MOS close to the system built on the clean counterpart.

Shan Yang, Lei Xie, Xiao Chen et al.

In this paper, we aim at improving the performance of synthesized speech in statistical parametric speech synthesis (SPSS) based on a generative adversarial network (GAN). In particular, we propose a novel architecture combining the traditional acoustic loss function and the GAN's discriminative loss under a multi-task learning (MTL) framework. The mean squared error (MSE) is usually used to estimate the parameters of deep neural networks, which only considers the numerical difference between the raw audio and the synthesized one. To mitigate this problem, we introduce the GAN as a second task to determine if the input is a natural speech with specific conditions. In this MTL framework, the MSE optimization improves the stability of GAN, and at the same time GAN produces samples with a distribution closer to natural speech. Listening tests show that the multi-task architecture can generate more natural speech that satisfies human perception than the conventional methods.