arXiv subjects

Wei Xue

Publications and source records attributed to Wei Xue.

At least 91 records · Page 5Linked to original sources

arXiv2025

Llasa+: Free Lunch for Accelerated and Streaming Llama-Based Speech Synthesis

Recent progress in text-to-speech (TTS) has achieved impressive naturalness and flexibility, especially with the development of large language model (LLM)-based approaches. However, existing autoregressive (AR) structures and large-scale models, such as Llasa, still face significant challenges in inference latency and streaming synthesis. To deal with the limitations, we introduce Llasa+, an accelerated and streaming TTS model built on Llasa. Specifically, to accelerate the generation process, we introduce two plug-and-play Multi-Token Prediction (MTP) modules following the frozen backbone. These modules allow the model to predict multiple tokens in one AR step. Additionally, to mitigate potential error propagation caused by inaccurate MTP, we design a novel verification algorithm that leverages the frozen backbone to validate the generated tokens, thus allowing Llasa+ to achieve speedup without sacrificing generation quality. Furthermore, we design a causal decoder that enables streaming speech reconstruction from tokens. Extensive experiments show that Llasa+ achieves a 1.48X speedup without sacrificing generation quality, despite being trained only on LibriTTS. Moreover, the MTP-and-verification framework can be applied to accelerate any LLM-based model. All codes and models are publicly available at https://github.com/ASLP-lab/LLaSA_Plus.

cs.SD↗

arXiv2025

Inference-time Scaling for Diffusion-based Audio Super-resolution

Diffusion models have demonstrated remarkable success in generative tasks, including audio super-resolution (SR). In many applications like movie post-production and album mastering, substantial computational budgets are available for achieving superior audio quality. However, while existing diffusion approaches typically increase sampling steps to improve quality, the performance remains fundamentally limited by the stochastic nature of the sampling process, leading to high-variance and quality-limited outputs. Here, rather than simply increasing the number of sampling steps, we propose a different paradigm through inference-time scaling for SR, which explores multiple solution trajectories during the sampling process. Different task-specific verifiers are developed, and two search algorithms, including the random search and zero-order search for SR, are introduced. By actively guiding the exploration of the high-dimensional solution space through verifier-algorithm combinations, we enable more robust and higher-quality outputs. Through extensive validation across diverse audio domains (speech, music, sound effects) and frequency ranges, we demonstrate consistent performance gains, achieving improvements of up to 9.70% in aesthetics, 5.88% in speaker similarity, 15.20% in word error rate, and 46.98% in spectral distance for speech SR from 4kHz to 24kHz, showcasing the effectiveness of our approach. Audio samples are available at: https://racerk.github.io/tt-scale-audiosr/.

cs.SD↗

arXiv2025

OW-CLIP: Data-Efficient Visual Supervision for Open-World Object Detection via Human-AI Collaboration

Open-world object detection (OWOD) extends traditional object detection to identifying both known and unknown object, necessitating continuous model adaptation as new annotations emerge. Current approaches face significant limitations: 1) data-hungry training due to reliance on a large number of crowdsourced annotations, 2) susceptibility to "partial feature overfitting," and 3) limited flexibility due to required model architecture modifications. To tackle these issues, we present OW-CLIP, a visual analytics system that provides curated data and enables data-efficient OWOD model incremental training. OW-CLIP implements plug-and-play multimodal prompt tuning tailored for OWOD settings and introduces a novel "Crop-Smoothing" technique to mitigate partial feature overfitting. To meet the data requirements for the training methodology, we propose dual-modal data refinement methods that leverage large language models and cross-modal similarity for data generation and filtering. Simultaneously, we develope a visualization interface that enables users to explore and deliver high-quality annotations: including class-specific visual feature phrases and fine-grained differentiated images. Quantitative evaluation demonstrates that OW-CLIP achieves competitive performance at 89% of state-of-the-art performance while requiring only 3.8% self-generated data, while outperforming SOTA approach when trained with equivalent data volumes. A case study shows the effectiveness of the developed method and the improved annotation quality of our visualization system.

cs.CV↗

arXiv2025

LEADRE: Multi-Faceted Knowledge Enhanced LLM Empowered Display Advertisement Recommender System

Display advertising provides significant value to advertisers, publishers, and users. Traditional display advertising systems utilize a multi-stage architecture consisting of retrieval, coarse ranking, and final ranking. However, conventional retrieval methods rely on ID-based learning to rank mechanisms and fail to adequately utilize the content information of ads, which hampers their ability to provide diverse recommendation lists. To address this limitation, we propose leveraging the extensive world knowledge of LLMs. However, three key challenges arise when attempting to maximize the effectiveness of LLMs: "How to capture user interests", "How to bridge the knowledge gap between LLMs and advertising system", and "How to efficiently deploy LLMs". To overcome these challenges, we introduce a novel LLM-based framework called LLM Empowered Display ADvertisement REcommender system (LEADRE). LEADRE consists of three core modules: (1) The Intent-Aware Prompt Engineering introduces multi-faceted knowledge and designs intent-aware pairs that fine-tune LLMs to generate ads tailored to users' personal interests. (2) The Advertising-Specific Knowledge Alignment incorporates auxiliary fine-tuning tasks and Direct Preference Optimization (DPO) to align LLMs with ad semantic and business value. (3) The Efficient System Deployment deploys LEADRE in an online environment by integrating both latency-tolerant and latency-sensitive service. Extensive offline experiments demonstrate the effectiveness of LEADRE and validate the contributions of individual modules. Online A/B test shows that LEADRE leads to a 1.57% and 1.17% GMV lift for serviced users on WeChat Channels and Moments separately. LEADRE has been deployed on both platforms, serving tens of billions of requests each day.

cs.IR↗

arXiv2025

MMStencil: Optimizing High-order Stencils on Multicore CPU using Matrix Unit

Matrix-accelerated stencil computation is a hot research topic, yet its application to three-dimensional (3D) high-order stencils and HPC remains underexplored. With the emergence of matrix units on multicore CPUs, we analyze matrix-based acceleration strategies and tailor an optimal approach for 3D high-order stencils. We introduce algorithmic optimizations based on SIMD and matrix units to address strided memory accesses, alignment conflicts, and redundant accesses. We propose memory optimizations to boost on-package memory efficiency, and a novel multi-thread parallelism paradigm to overcome data-sharing challenges caused by the absence of shared data caches. MMStencil sustains consistently high hardware utilization across diverse stencil shapes and dimensions. Our DMA-based inter-NUMA communication further mitigates NUMA effects and MPI limitations in hybrid parallelism. Combining all the innovations, MMStencil outperforms state-of-the-art libraries on Nvidia A100 GPGPU by up to 2.1x. Moreover, the performance improvements translate directly to real-world HPC applications and enable RTM applications to yield 1.8x speedup versus a highly optimized industrial Nvidia A100 GPGPU version.

cs.DC↗

arXiv2025

CMD: Controllable Multiview Diffusion for 3D Editing and Progressive Generation

Recently, 3D generation methods have shown their powerful ability to automate 3D model creation. However, most 3D generation methods only rely on an input image or a text prompt to generate a 3D model, which lacks the control of each component of the generated 3D model. Any modifications of the input image lead to an entire regeneration of the 3D models. In this paper, we introduce a new method called CMD that generates a 3D model from an input image while enabling flexible local editing of each component of the 3D model. In CMD, we formulate the 3D generation as a conditional multiview diffusion model, which takes the existing or known parts as conditions and generates the edited or added components. This conditional multiview diffusion model not only allows the generation of 3D models part by part but also enables local editing of 3D models according to the local revision of the input image without changing other 3D parts. Extensive experiments are conducted to demonstrate that CMD decomposes a complex 3D generation task into multiple components, improving the generation quality. Meanwhile, CMD enables efficient and flexible local editing of a 3D model by just editing one rendered image.

cs.CV↗

arXiv2025

StructMG: A Fast and Scalable Structured Algebraic Multigrid

Parallel multigrid is widely used as preconditioners in solving large-scale sparse linear systems. However, the current multigrid library still needs more satisfactory performance for structured grid problems regarding speed and scalability. Based on the classical 'multigrid seesaw', we derive three necessary principles for an efficient structured multigrid, which instructs our design and implementation of StructMG, a fast and scalable algebraic multigrid that constructs hierarchical grids automatically. As a preconditioner, StructMG can achieve both low cost per iteration and good convergence when solving large-scale linear systems with iterative methods in parallel. A stencil-based triple-matrix product via symbolic derivation and code generation is proposed for multi-dimensional Galerkin coarsening to reduce grid complexity, operator complexity, and implementation effort. A unified parallel framework of sparse triangular solver is presented to achieve fast convergence and high parallel efficiency for smoothers, including dependence-preserving Gauss-Seidel and incomplete LU methods. Idealized and real-world problems from radiation hydrodynamics, petroleum reservoir simulation, numerical weather prediction, and solid mechanics, are evaluated on ARM and X86 platforms to show StructMG's effectiveness. In comparison to \textit{hypre}'s structured and general multigrid preconditioners, StructMG achieves the fastest time-to-solutions in all cases with average speedups of 15.5x, 5.5x, 6.7x, 7.3x over SMG, PFMG, SysPFMG, and BoomerAMG, respectively. StructMG also significantly improves strong and weak scaling efficiencies.

math.NA↗

arXiv2025

EVA: An Embodied World Model for Future Video Anticipation

Video generation models have made significant progress in simulating future states, showcasing their potential as world simulators in embodied scenarios. However, existing models often lack robust understanding, limiting their ability to perform multi-step predictions or handle Out-of-Distribution (OOD) scenarios. To address this challenge, we propose the Reflection of Generation (RoG), a set of intermediate reasoning strategies designed to enhance video prediction. It leverages the complementary strengths of pre-trained vision-language and video generation models, enabling them to function as a world model in embodied scenarios. To support RoG, we introduce Embodied Video Anticipation Benchmark(EVA-Bench), a comprehensive benchmark that evaluates embodied world models across diverse tasks and scenarios, utilizing both in-domain and OOD datasets. Building on this foundation, we devise a world model, Embodied Video Anticipator (EVA), that follows a multistage training paradigm to generate high-fidelity video frames and apply an autoregressive strategy to enable adaptive generalization for longer video sequences. Extensive experiments demonstrate the efficacy of EVA in various downstream tasks like video generation and robotics, thereby paving the way for large-scale pre-trained models in real-world video prediction applications. The video demos are available at \hyperlink{https://sites.google.com/view/icml-eva}{https://sites.google.com/view/icml-eva}.

cs.CV↗

arXiv2025

LLaSE-G1: Incentivizing Generalization Capability for LLaMA-based Speech Enhancement

Recent advancements in language models (LMs) have demonstrated strong capabilities in semantic understanding and contextual modeling, which have flourished in generative speech enhancement (SE). However, many LM-based SE approaches primarily focus on semantic information, often neglecting the critical role of acoustic information, which leads to acoustic inconsistency after enhancement and limited generalization across diverse SE tasks. In this paper, we introduce LLaSE-G1, a LLaMA-based language model that incentivizes generalization capabilities for speech enhancement. LLaSE-G1 offers the following key contributions: First, to mitigate acoustic inconsistency, LLaSE-G1 employs continuous representations from WavLM as input and predicts speech tokens from X-Codec2, maximizing acoustic preservation. Second, to promote generalization capability, LLaSE-G1 introduces dual-channel inputs and outputs, unifying multiple SE tasks without requiring task-specific IDs. Third, LLaSE-G1 outperforms prior task-specific discriminative and generative SE models, demonstrating scaling effects at test time and emerging capabilities for unseen SE tasks. Additionally, we release our code and models to support further research in this area.

eess.AS↗

arXiv2025

FlashAudio: Rectified Flows for Fast and High-Fidelity Text-to-Audio Generation

Recent advancements in latent diffusion models (LDMs) have markedly enhanced text-to-audio generation, yet their iterative sampling processes impose substantial computational demands, limiting practical deployment. While recent methods utilizing consistency-based distillation aim to achieve few-step or single-step inference, their one-step performance is constrained by curved trajectories, preventing them from surpassing traditional diffusion models. In this work, we introduce FlashAudio with rectified flows to learn straight flow for fast simulation. To alleviate the inefficient timesteps allocation and suboptimal distribution of noise, FlashAudio optimizes the time distribution of rectified flow with Bifocal Samplers and proposes immiscible flow to minimize the total distance of data-noise pairs in a batch vias assignment. Furthermore, to address the amplified accumulation error caused by the classifier-free guidance (CFG), we propose Anchored Optimization, which refines the guidance scale by anchoring it to a reference trajectory. Experimental results on text-to-audio generation demonstrate that FlashAudio's one-step generation performance surpasses the diffusion-based models with hundreds of sampling steps on audio quality and enables a sampling speed of 400x faster than real-time on a single NVIDIA 4090Ti GPU. Code will be available at https://github.com/liuhuadai/FlashAudio.

eess.AS↗

arXiv2025

OmniAudio: Generating Spatial Audio from 360-Degree Video

Traditional video-to-audio generation techniques primarily focus on perspective video and non-spatial audio, often missing the spatial cues necessary for accurately representing sound sources in 3D environments. To address this limitation, we introduce a novel task, 360V2SA, to generate spatial audio from 360-degree videos, specifically producing First-order Ambisonics (FOA) audio - a standard format for representing 3D spatial audio that captures sound directionality and enables realistic 3D audio reproduction. We first create Sphere360, a novel dataset tailored for this task that is curated from real-world data. We also design an efficient semi-automated pipeline for collecting and cleaning paired video-audio data. To generate spatial audio from 360-degree video, we propose a novel framework OmniAudio, which leverages self-supervised pre-training using both spatial audio data (in FOA format) and large-scale non-spatial data. Furthermore, OmniAudio features a dual-branch framework that utilizes both panoramic and perspective video inputs to capture comprehensive local and global information from 360-degree videos. Experimental results demonstrate that OmniAudio achieves state-of-the-art performance across both objective and subjective metrics on Sphere360. Code and datasets are available at https://github.com/liuhuadai/OmniAudio. The project website is available at https://OmniAudio-360V2SA.github.io.

eess.AS↗

arXiv2025

It's Not a Walk in the Park! Challenges of Idiom Translation in Speech-to-text Systems

Idioms are defined as a group of words with a figurative meaning not deducible from their individual components. Although modern machine translation systems have made remarkable progress, translating idioms remains a major challenge, especially for speech-to-text systems, where research on this topic is notably sparse. In this paper, we systematically evaluate idiom translation as compared to conventional news translation in both text-to-text machine translation (MT) and speech-to-text translation (SLT) systems across two language pairs (German to English, Russian to English). We compare state-of-the-art end-to-end SLT systems (SeamlessM4T SLT-to-text, Whisper Large v3) with MT systems (SeamlessM4T SLT-to-text, No Language Left Behind), Large Language Models (DeepSeek, LLaMA) and cascaded alternatives. Our results reveal that SLT systems experience a pronounced performance drop on idiomatic data, often reverting to literal translations even in higher layers, whereas MT systems and Large Language Models demonstrate better handling of idioms. These findings underscore the need for idiom-specific strategies and improved internal representations in SLT architectures.

cs.CL↗

arXiv2025

MMAR: A Challenging Benchmark for Deep Reasoning in Speech, Audio, Music, and Their Mix

We introduce MMAR, a new benchmark designed to evaluate the deep reasoning capabilities of Audio-Language Models (ALMs) across massive multi-disciplinary tasks. MMAR comprises 1,000 meticulously curated audio-question-answer triplets, collected from real-world internet videos and refined through iterative error corrections and quality checks to ensure high quality. Unlike existing benchmarks that are limited to specific domains of sound, music, or speech, MMAR extends them to a broad spectrum of real-world audio scenarios, including mixed-modality combinations of sound, music, and speech. Each question in MMAR is hierarchically categorized across four reasoning layers: Signal, Perception, Semantic, and Cultural, with additional sub-categories within each layer to reflect task diversity and complexity. To further foster research in this area, we annotate every question with a Chain-of-Thought (CoT) rationale to promote future advancements in audio reasoning. Each item in the benchmark demands multi-step deep reasoning beyond surface-level understanding. Moreover, a part of the questions requires graduate-level perceptual and domain-specific knowledge, elevating the benchmark's difficulty and depth. We evaluate MMAR using a broad set of models, including Large Audio-Language Models (LALMs), Large Audio Reasoning Models (LARMs), Omni Language Models (OLMs), Large Language Models (LLMs), and Large Reasoning Models (LRMs), with audio caption inputs. The performance of these models on MMAR highlights the benchmark's challenging nature, and our analysis further reveals critical limitations of understanding and reasoning capabilities among current models. We hope MMAR will serve as a catalyst for future advances in this important but little-explored area.

cs.SD↗

arXiv2025

J1: Exploring Simple Test-Time Scaling for LLM-as-a-Judge

The current focus of AI research is shifting from emphasizing model training towards enhancing evaluation quality, a transition that is crucial for driving further advancements in AI systems. Traditional evaluation methods typically rely on reward models assigning scalar preference scores to outputs. Although effective, such approaches lack interpretability, leaving users often uncertain about why a reward model rates a particular response as high or low. The advent of LLM-as-a-Judge provides a more scalable and interpretable method of supervision, offering insights into the decision-making process. Moreover, with the emergence of large reasoning models, which consume more tokens for deeper thinking and answer refinement, scaling test-time computation in the LLM-as-a-Judge paradigm presents an avenue for further boosting performance and providing more interpretability through reasoning traces. In this paper, we introduce $\textbf{J1-7B}$, which is first supervised fine-tuned on reflection-enhanced datasets collected via rejection-sampling and subsequently trained using Reinforcement Learning (RL) with verifiable rewards. At inference time, we apply Simple Test-Time Scaling (STTS) strategies for additional performance improvement. Experimental results demonstrate that $\textbf{J1-7B}$ surpasses the previous state-of-the-art LLM-as-a-Judge by $ \textbf{4.8}$\% and exhibits a $ \textbf{5.1}$\% stronger scaling trend under STTS. Additionally, we present three key findings: (1) Existing LLM-as-a-Judge does not inherently exhibit such scaling trend. (2) Model simply fine-tuned on reflection-enhanced datasets continues to demonstrate similarly weak scaling behavior. (3) Significant scaling trend emerges primarily during the RL phase, suggesting that effective STTS capability is acquired predominantly through RL training.

cs.LG↗

arXiv2025

SongEval: A Benchmark Dataset for Song Aesthetics Evaluation

Aesthetics serve as an implicit and important criterion in song generation tasks that reflect human perception beyond objective metrics. However, evaluating the aesthetics of generated songs remains a fundamental challenge, as the appreciation of music is highly subjective. Existing evaluation metrics, such as embedding-based distances, are limited in reflecting the subjective and perceptual aspects that define musical appeal. To address this issue, we introduce SongEval, the first open-source, large-scale benchmark dataset for evaluating the aesthetics of full-length songs. SongEval includes over 2,399 songs in full length, summing up to more than 140 hours, with aesthetic ratings from 16 professional annotators with musical backgrounds. Each song is evaluated across five key dimensions: overall coherence, memorability, naturalness of vocal breathing and phrasing, clarity of song structure, and overall musicality. The dataset covers both English and Chinese songs, spanning nine mainstream genres. Moreover, to assess the effectiveness of song aesthetic evaluation, we conduct experiments using SongEval to predict aesthetic scores and demonstrate better performance than existing objective evaluation metrics in predicting human-perceived musical quality.

eess.AS↗

arXiv2025

VidMuse: A Simple Video-to-Music Generation Framework with Long-Short-Term Modeling

In this work, we systematically study music generation conditioned solely on the video. First, we present a large-scale dataset comprising 360K video-music pairs, including various genres such as movie trailers, advertisements, and documentaries. Furthermore, we propose VidMuse, a simple framework for generating music aligned with video inputs. VidMuse stands out by producing high-fidelity music that is both acoustically and semantically aligned with the video. By incorporating local and global visual cues, VidMuse enables the creation of musically coherent audio tracks that consistently match the video content through Long-Short-Term modeling. Through extensive experiments, VidMuse outperforms existing models in terms of audio quality, diversity, and audio-visual alignment. The code and datasets are available at https://vidmuse.github.io/.

cs.CV↗

arXiv2025

Co$^{3}$Gesture: Towards Coherent Concurrent Co-speech 3D Gesture Generation with Interactive Diffusion

Generating gestures from human speech has gained tremendous progress in animating virtual avatars. While the existing methods enable synthesizing gestures cooperated by individual self-talking, they overlook the practicality of concurrent gesture modeling with two-person interactive conversations. Moreover, the lack of high-quality datasets with concurrent co-speech gestures also limits handling this issue. To fulfill this goal, we first construct a large-scale concurrent co-speech gesture dataset that contains more than 7M frames for diverse two-person interactive posture sequences, dubbed GES-Inter. Additionally, we propose Co$^3$Gesture, a novel framework that enables coherent concurrent co-speech gesture synthesis including two-person interactive movements. Considering the asymmetric body dynamics of two speakers, our framework is built upon two cooperative generation branches conditioned on separated speaker audio. Specifically, to enhance the coordination of human postures with respect to corresponding speaker audios while interacting with the conversational partner, we present a Temporal Interaction Module (TIM). TIM can effectively model the temporal association representation between two speakers' gesture sequences as interaction guidance and fuse it into the concurrent gesture generation. Then, we devise a mutual attention mechanism to further holistically boost learning dependencies of interacted concurrent motions, thereby enabling us to generate vivid and coherent gestures. Extensive experiments demonstrate that our method outperforms the state-of-the-art models on our newly collected GES-Inter dataset. The dataset and source code are publicly available at \href{https://mattie-e.github.io/Co3/}{\textit{https://mattie-e.github.io/Co3/}}.

cs.CV↗

arXiv2025

DB-GNN: Dual-Branch Graph Neural Network with Multi-Level Contrastive Learning for Jointly Identifying Within- and Cross-Frequency Coupled Brain Networks

Within-frequency coupling (WFC) and cross-frequency coupling (CFC) in brain networks reflect neural synchronization within the same frequency band and cross-band oscillatory interactions, respectively. Their synergy provides a comprehensive understanding of neural mechanisms underlying cognitive states such as emotion. However, existing multi-channel EEG studies often analyze WFC or CFC separately, failing to fully leverage their complementary properties. This study proposes a dual-branch graph neural network (DB-GNN) to jointly identify within- and cross-frequency coupled brain networks. Firstly, DBGNN leverages its unique dual-branch learning architecture to efficiently mine global collaborative information and local cross-frequency and within-frequency coupling information. Secondly, to more fully perceive the global information of cross-frequency and within-frequency coupling, the global perception branch of DB-GNN adopts a Transformer architecture. To prevent overfitting of the Transformer architecture, this study integrates prior within- and cross-frequency coupling information into the Transformer inference process, thereby enhancing the generalization capability of DB-GNN. Finally, a multi-scale graph contrastive learning regularization term is introduced to constrain the global and local perception branches of DB-GNN at both graph-level and node-level, enhancing its joint perception ability and further improving its generalization performance. Experimental validation on the emotion recognition dataset shows that DB-GNN achieves a testing accuracy of 97.88% and an F1- score of 97.87%, reaching the state-of-the-art performance.

q-bio.NC↗