arXiv subjects

Ziyue Jiang

Publications and source records attributed to Ziyue Jiang.

At least 19 recordsLinked to original sources

arXiv2026

UniVoice: Unifying Autoregressive ASR and Flow-Matching based TTS with Large Language Models

Large language models (LLMs) have demonstrated promising performance in both automatic speech recognition (ASR) and text-to-speech (TTS) systems, gradually becoming the mainstream approach. However, most current approaches address these tasks separately rather than through a unified framework. This work aims to integrate these two tasks into one unified model. Although discrete speech tokenization enables joint modeling, its inherent information loss limits performance in both recognition and generation. In this work, we present UniVoice, a unified LLM framework through continuous representations that seamlessly integrates speech recognition and synthesis within a single model. Our approach combines the strengths of autoregressive modeling for speech recognition with flow matching for high-quality generation. To mitigate the inherent divergence between autoregressive and flow-matching models, we further design a dual attention mechanism, which switches between a causal mask for recognition and a bidirectional attention mask for synthesis. Furthermore, the proposed text-prefix-conditioned speech infilling method enables high-fidelity zero-shot voice cloning. Experimental results demonstrate that our method can achieve or exceed current single-task modeling methods in both ASR and zero-shot TTS tasks. This work explores new possibilities for end-to-end speech understanding and generation. Code is available at https://github.com/gwh22/UniVoice.

eess.AS↗

arXiv2026

OmniJudge or OmniBias? Diagnosing Multimodal Judges through Balanced, Decoupled Lenses

Multimodal understanding models that can jointly judge text-to-image (T2I), text-to-video (T2V) and text-to-speech (TTS) generation are increasingly used as "OmniJudges" for evaluation and automatic annotation. How reliably they understand what they score remains unclear, since existing benchmarks and training data tend to overemphasize positive examples and to conflate distinct failure modes, so a judge may score well without recognizing failures while its capability gaps stay hidden. Motivated by this, we introduce D3-Omni, a balanced and decoupled benchmark for diagnosing fine-grained multimodal understanding, covering 53 orthogonal binary dimensions (17/22/14) and 10,671 samples (3,526/1,998/5,147) across the three tasks. Rather than re-generating outputs, which may leak information across dimensions, we fix verified fully positive seeds and derive negatives through controlled prompt rewriting and atomic, dimension-isolating perturbations. The resulting D3 design is Dual-balanced, which helps alleviate negative-sample scarcity and per-dimension label imbalance; Decoupled, so that each error is attributable to a single capability; and Dynamic, steering construction toward under-represented regions of the label distribution as generative models improve.The suite reaches near 1:1 per-dimension parity and a uniform distribution over all total-score levels. Under this balanced view, even strong OmniJudges tend to struggle on modality-related dimensions, to confirm satisfied requirements far more reliably than they detect violated ones, and to treat nominally distinct attributes as largely a single decision, suggesting that aggregate accuracy may hide systematic blind spots that a balanced and decoupled lens can help expose and, in turn, address.

cs.AI↗

arXiv2026

VoxAudio: Vocalized Audio Synthesis via Multi-Reward Autoregressive Flow Matching

Vocalized audio synthesis, the task of generating audio in which intelligible speech is embedded within an environmental soundscape, underpins applications such as podcast production and video dubbing. Existing Text-to-Audio (T2A) systems either reduce quoted speech to unintelligible vocal murmur or delegate it to a separate TTS model with post-hoc mixing, which forfeits control over when speech occurs and how it interacts with the scene. We present VoxAudio, a causal autoregressive flow matching model that addresses this problem from three complementary aspects. At the architecture level, chunk-wise causal factorization with independent per-chunk noise levels lets audio be emitted through sliding-window streaming inference with KV caching at variable target durations; to enable inference at arbitrary chunk granularities, we further pretrain the model with randomized chunk boundaries. At the preference level, multi-reward Negative-aware FineTuning (NFT) jointly optimizes semantic fidelity, linguistic accuracy, aesthetic quality, and temporal grounding At the data level, to supply the missing supervision for vocal content, we build VoxCorpus, a large-scale corpus whose captions quote the verbatim transcript of embedded speech with time intervals, and VoxBench, an interval-annotated benchmark with a temporal-grounding metric. Experiments on four benchmarks spanning general audio, speech, and unified vocalized audio validate the effectiveness and efficiency of VoxAudio. Our code and demos are available at https://voxaudio.github.io.

cs.SD↗

arXiv2026

Qwen-Audio-VAE Technical Report

We introduce \textbf{Qwen-Audio-VAE}, a suite of low-bitrate, fast-encoding continuous audio autoencoders designed for scalable general audio generation. The model is built around a simple but important principle: an audio VAE should not only reconstruct diverse audio with high fidelity, but also produce compact latent representations fast enough to support large-scale text-to-audio training. Qwen-Audio-VAE combines a causal encoder-decoder, window Transformer blocks, and multi-discriminator training to achieve a strong balance between reconstruction quality and compression rate. The model is trained at scale on 5 million hours of multi-domain audio, enabling robust reconstruction across heterogeneous acoustic conditions. To further improve computational efficiency, we adopt an asymmetric encoder-decoder backbone and introduce latency-aware encoder pruning to maximize encoding throughput. Experiments on public speech, music, and sound reconstruction benchmarks show that Qwen-Audio-VAE generalizes well across diverse audio domains and is particularly efficient, requiring only 541 ms to encode 32 minutes of audio. Overall, Qwen-Audio-VAE provides a high-quality, compact, and high-throughput representation backbone for efficient general audio generation.

eess.AS↗

arXiv2026

Agentic RAG-VLM: Affordance-Aware Retrieval-Augmented Generation with Self-Reflective Planning for Robotic Grasping

Generalizable robotic grasping in cluttered environments is essential for deploying manipulators in unstructured human spaces, yet existing VLM-based methods rely on visual similarity for object matching, neglecting physical affordances such as handle graspability and material fragility, and operate open-loop without spatial reasoning or failure recovery, limiting their effectiveness when objects are densely packed or physically diverse. We present Agentic RAG-VLM, a unified framework that bridges VLM-based semantic understanding and physically grounded grasp execution by integrating retrieval-augmented generation (RAG) with vision-language models (VLMs) and agentic self-reflective planning. Agentic RAG-VLM introduces three tightly coupled components: (1) a Hierarchical Affordance-Aware RAG (HAA-RAG) that encodes four-dimensional affordance descriptors, including type, material, fragility, and graspable region, and retrieves strategies by functional affordance compatibility rather than visual appearance; (2) a Scene Graph Constraint Reasoner that constructs spatial relationship graphs from VLM perception and translates proximity, occlusion, and support constraints into concrete grasp parameter adjustments; and (3) an Agentic Self-Reflective Pipeline with a 14-type failure taxonomy and three-level adaptive retry for closed-loop grasp refinement. Evaluated on a 12-task benchmark spanning single-grasp, interactive, and long-horizon scenarios with 360 trials per configuration, Agentic RAG-VLM achieves 78.3 percent overall success, a 53.3 percentage-point absolute gain over VLM-only baselines, demonstrating that affordance-aware retrieval, scene graph reasoning, and agentic recovery are jointly essential for robust manipulation.

cs.AI↗

arXiv2026

A Survey of Full-Duplex Spoken Dialogue Systems: Architectural Hierarchy, Interaction Ontology, and Decision State Machine

More than a dozen spoken dialogue systems have recently claimed to be "full-duplex," yet the term has been used to describe substantially different capabilities. Existing surveys collapse them onto a single axis (cascaded/end-to-end, or engineered/learned) and miss the distinctions that matter most for builders. We argue that much of this ambiguity is taxonomical: current terminology does not specify where duplex decisions are made, which interaction types are supported, or how a system behaves moment by moment. This paper introduces three complementary frameworks: (i) an L0-L3 Architectural Hierarchy that locates where duplex decisions are made; (ii) a $T\times I\times R$ Interaction Ontology that specifies the temporal relation, user intent, and required system response for each interaction; and (iii) a Decision State Machine (IDLE/LISTEN/SPEAK/WAIT/DUAL) that describes how systems move between states. Across published systems and benchmarks, our audit documents a realization gap: although many architectures can in principle operate in full-duplex states, their observed behavior remains constrained by the interaction patterns represented in training and evaluation. We point to the limited public training-data coverage relative to the (largely undisclosed) industrial corpora, together with the still-unrealized goal of L3 representation-level modeling, as the key frontiers for future research on full-duplex dialogue. The related material is available at https://github.com/DuplexLM/DuplexSurvey.

eess.AS↗

arXiv2026

Comprehensive Benchmarking of Long-Form Speech Generation in Diverse Scenarios

Recent advances in speech generation have enabled high-fidelity synthesis, yet systematic evaluation of models under long-context conditions remains largely underexplored. A comprehensive evaluation benchmark for long-form speech is indispensable for two reasons: 1) existing test scenarios are often confined to limited domains, creating a significant gap with the diverse downstream applications; 2) existing metrics overlook critical long-text factors such as consistency and coherence, failing to generalize reliably. To this end, we propose Swanbench-Speech, a comprehensive benchmark that decomposes long-form speech quality into specific, disentangled dimensions. SwanBench-Speech has three key properties. 1) Rich speech scenarios: Focusing on long-form speech generation and dialog generation, SwanBench-Speech covers acoustics, semantics, and expressiveness challenges, and consists of 1,101 samples spanning 17 common speech scenarios; 2) Comprehensive evaluation dimensions: Along the acoustics, semantics, and expressiveness axes, SwanBench-Speech defines an automated evaluation protocol with seven metrics to provide a comprehensive, accurate, and standardized assessment; 3) Valuable Insights: Through extensive experiments, we reveal that current models still struggle in highly expressive scenarios and exhibit a notable gap in consistency and hierarchy compared to real recordings.

eess.AS↗

arXiv2026

Discovery and Characterization of White Dwarf-FGK Main-Sequence Binaries within the Optical Main-Sequence Locus

White dwarf main-sequence (WDMS) binaries provide important laboratories for studying binary evolution and the formation of low-mass white dwarfs. In this work, we identify 654 reliable WDMS candidates with FGK-type companions from an initial set of 772 ultraviolet-excess sources, selected using stellar atmospheric parameters from LAMOST spectroscopy and subsequently refined with \textit{Gaia} DR3 astrometry and photometry together with ultraviolet data from \textit{GALEX}. Candidates were selected based on ultraviolet excess relative to the \textit{Gaia} main-sequence locus and refined using isochrone constraints to exclude systems inconsistent with MS companions. Binary spectral energy distribution fitting yields effective temperatures and radii for both components, as well as distance and extinction estimates. The MS companions are dominated by G-type stars ($\sim52\%$), with comparable fractions of F- and K-type companions, and no A-type primaries. Using white-dwarf evolutionary cooling models, we find that the WD components are predominantly low-mass ($M_{\rm WD}\,\sim\,0.2\text{--}0.4\,M_\odot$), including a substantial population of extremely low-mass ($<0.3\,M_\odot$) WDs likely produced through binary interaction. The WDs are generally hot ($\sim1.5\times10^4\,\mathrm{K}$), consistent with the ultraviolet selection bias favoring luminous, large-radius WDs. Multi-epoch LAMOST radial velocities show larger amplitudes than those of a comparison sample of MS stars, supporting the close-binary nature of these systems. Although subject to strong selection effects, the catalog offers a clean and well-characterized sample of FGK+WD binaries.

astro-ph.SR↗

arXiv2026

Revisiting Ca II Activity Indices in FGK Stars: Systematic Biases in Infrared Triplet Measurements

Synthetic-template subtraction is widely used to measure chromospheric activity in large spectroscopic surveys. However, many solar-like FGK stars show systematically negative Ca II infrared triplet (IRT) residual indices, implying that the observed line cores are deeper than those predicted by parameter-matched templates. We investigate this effect using solar-like stars from LAMOST DR9, MaStar, and XSL DR3, measuring activity indices (R+) for both the Ca II H&K and IRT lines in a uniform framework. We find that observational effects, including atmospheric-parameter offsets, treatment of the instrumental line-spread function, and propagated measurement uncertainties, contribute to scatter but do not explain the systematic negative bias in R+_IRT. The results instead suggest that the negative bias most likely arises because photospheric templates underestimate the depth of the IRT cores, likely owing to missing chromospheric structure and, to a lesser extent, NLTE effects. An empirical increase in the adopted microturbulent velocity deepens the synthetic IRT cores and partially mitigates the negative offset. In addition, R+ values derived from different synthesis configurations show systematic offsets but generally preserve strong linear correlations, indicating that they can be cross-calibrated. These results clarify the origin of negative Ca II IRT residual indices and help interpret template-dependent systematics in chromospheric activity measurements based on synthetic-template subtraction.

astro-ph.SR↗

arXiv2026

Characterizing the Instrumental Profile of LAMOST

The instrumental profile (IP) of a telescope is of great significance for spectroscopic analyses, especially for wavelength calibration and stellar parameter measurements. The Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST) employs arc lamps for wavelength calibration. These lamps produce sharp emission lines with known wavelengths, and the observed arc lamp spectra can well characterize the IP. However, IPs are influenced by multiple factors, making them difficult to model accurately with traditional methods. Neural networks, which can automatically capture complex patterns and nonlinear features in data, provide a promising approach for high-precision IP measurement. We therefore construct a multi-layer perceptron (MLP) based on The Payne neural network to derive IPs for LAMOST. After training, the model can retrieve the IP for any fiber, at any wavelength, and at any time. We then apply the derived IP to stellar radial velocity (RV) measurements and analyze the impact of different IP center localization methods on the results. Finally, the dispersion of the measured RVs is reduced by approximately 3 km/s. This improvement will facilitate the search for long-period binary stars via RV variations.

astro-ph.IM↗

arXiv2026

Qwen3-TTS Technical Report

In this report, we present the Qwen3-TTS series, a family of advanced multilingual, controllable, robust, and streaming text-to-speech models. Qwen3-TTS supports state-of-the-art 3-second voice cloning and description-based control, allowing both the creation of entirely novel voices and fine-grained manipulation over the output speech. Trained on over 5 million hours of speech data spanning 10 languages, Qwen3-TTS adopts a dual-track LM architecture for real-time synthesis, coupled with two speech tokenizers: 1) Qwen-TTS-Tokenizer-25Hz is a single-codebook codec emphasizing semantic content, which offers seamlessly integration with Qwen-Audio and enables streaming waveform reconstruction via a block-wise DiT. 2) Qwen-TTS-Tokenizer-12Hz achieves extreme bitrate reduction and ultra-low-latency streaming, enabling immediate first-packet emission ($97\,\mathrm{ms}$) through its 12.5 Hz, 16-layer multi-codebook design and a lightweight causal ConvNet. Extensive experiments indicate state-of-the-art performance across diverse objective and subjective benchmark (e.g., TTS multilingual test set, InstructTTSEval, and our long speech test set). To facilitate community research and development, we release both tokenizers and models under the Apache 2.0 license.

cs.SD↗

arXiv2025

Missing Data Imputation by Reducing Mutual Information with Rectified Flows

This paper introduces a novel iterative method for missing data imputation that sequentially reduces the mutual information between data and the corresponding missingness mask. Inspired by GAN-based approaches that train generators to decrease the predictability of missingness patterns, our method explicitly targets this reduction in mutual information. Specifically, our algorithm iteratively minimizes the KL divergence between the joint distribution of the imputed data and missingness mask, and the product of their marginals from the previous iteration. We show that the optimal imputation under this framework can be achieved by solving an ODE whose velocity field minimizes a rectified flow training objective. We further illustrate that some existing imputation techniques can be interpreted as approximate special cases of our mutual-information-reducing framework. Comprehensive experiments on synthetic and real-world datasets validate the efficacy of our proposed approach, demonstrating its superior imputation performance. Our implementation is available at https://github.com/yujhml/MIRI-Imputation.

stat.ML↗

arXiv2025

WavReward: Spoken Dialogue Models With Generalist Reward Evaluators

End-to-end spoken dialogue models such as GPT-4o-audio have recently garnered significant attention in the speech domain. However, the evaluation of spoken dialogue models' conversational performance has largely been overlooked. This is primarily due to the intelligent chatbots convey a wealth of non-textual information which cannot be easily measured using text-based language models like ChatGPT. To address this gap, we propose WavReward, a reward feedback model based on audio language models that can evaluate both the IQ and EQ of spoken dialogue systems with speech input. Specifically, 1) based on audio language models, WavReward incorporates the deep reasoning process and the nonlinear reward mechanism for post-training. By utilizing multi-sample feedback via the reinforcement learning algorithm, we construct a specialized evaluator tailored to spoken dialogue models. 2) We introduce ChatReward-30K, a preference dataset used to train WavReward. ChatReward-30K includes both comprehension and generation aspects of spoken dialogue models. These scenarios span various tasks, such as text-based chats, nine acoustic attributes of instruction chats, and implicit chats. WavReward outperforms previous state-of-the-art evaluation models across multiple spoken dialogue scenarios, achieving a substantial improvement about Qwen2.5-Omni in objective accuracy from 53.4$\%$ to 91.5$\%$. In subjective A/B testing, WavReward also leads by a margin of 83$\%$. Comprehensive ablation studies confirm the necessity of each component of WavReward. All data and code will be publicly at https://github.com/jishengpeng/WavReward after the paper is accepted.

eess.AS↗

arXiv2025

DiTReducio: A Training-Free Acceleration for DiT-Based TTS via Progressive Calibration

While Diffusion Transformers (DiT) have advanced non-autoregressive (NAR) speech synthesis, their high computational demands remain an limitation. Existing DiT-based text-to-speech (TTS) model acceleration approaches mainly focus on reducing sampling steps through distillation techniques, yet they remain constrained by training costs. We introduce DiTReducio, a training-free acceleration framework that compresses computations in DiT-based TTS models via progressive calibration. We propose two compression methods, Temporal Skipping and Branch Skipping, to eliminate redundant computations during inference. Moreover, based on two characteristic attention patterns identified within DiT layers, we devise a pattern-guided strategy to selectively apply the compression methods. Our method allows flexible modulation between generation quality and computational efficiency through adjustable compression thresholds. Experimental evaluations conducted on F5-TTS and MegaTTS 3 demonstrate that DiTReducio achieves a 75.4% reduction in FLOPs and improves the Real-Time Factor (RTF) by 37.1%, while preserving generation quality.

cs.SD↗

arXiv2025

The survey of DA double white dwarf candidates based on DESI EDR

Mergers of double white dwarfs are considered significant potential progenitors of type Ia supernovae. Although there is no direct observational evidence to definitively determine the formation pathways of SNe Ia, studying the physical properties of DWDs provides valuable insights into their evolutionary processes, interaction modes, and merger mechanisms, which are essential for understanding the explosion mechanisms of SNe Ia. This study aims to identify DWD candidates through spectroscopic radial velocity measurements and analyze their physical properties based on DESI EDR. We crossmatched DESI EDR with the Gaia EDR3 to select DA. We measured the spectroscopic RV using the cross-correlation function and assessed the significance of RV variability using a chi-squared-based variability method. Spectroscopic Teff and log g were derived by fitting the hydrogen Balmer lines, with 3D convection corrections applied. Orbital periods and semi-amplitudes were obtained through a Lomb-Scargle analysis of the RV time series. We interpolated WD cooling models and applied Monte Carlo simulations to calculate masses, cooling ages, radii, and their associated uncertainties. We also analyzed their photometric and spectral energy distribution properties to derive photometric temperatures and radii, which were then compared with the corresponding spectroscopic parameters. We identified 33 DA DWD candidates with significant RV variability, including 28 new discoveries. Among them, we found an extremely low-mass DWD candidate and a potential triple system. For these candidates, we measured key physical parameters including Teff, log g, mass, and radius, and estimated the orbital periods based on the available data. Of these, 17 candidates exhibit relatively clear periodic RV variability in the current data, and we report their best-fitting periods and RV semi-amplitudes.

astro-ph.SR↗

arXiv2025

Versatile Framework for Song Generation with Prompt-based Control

Song generation focuses on producing controllable high-quality songs based on various prompts. However, existing methods struggle to generate vocals and accompaniments with prompt-based control and proper alignment. Additionally, they fall short in supporting various tasks. To address these challenges, we introduce VersBand, a multi-task song generation framework for synthesizing high-quality, aligned songs with prompt-based control. VersBand comprises these primary models: 1) VocalBand, a decoupled model, leverages the flow-matching method for generating singing styles, pitches, and mel-spectrograms, allowing fast, high-quality vocal generation with style control. 2) AccompBand, a flow-based transformer model, incorporates the Band-MOE, selecting suitable experts for enhanced quality, alignment, and control. This model allows for generating controllable, high-quality accompaniments aligned with vocals. 3) Two generation models, LyricBand for lyrics and MelodyBand for melodies, contribute to the comprehensive multi-task song generation system, allowing for extensive control based on multiple prompts. Experimental results show that VersBand outperforms baseline models across multiple song generation tasks using objective and subjective metrics. Demos and codes are available at https://aaronz345.github.io/VersBandDemo and https://github.com/AaronZ345/VersBand.

eess.AS↗

arXiv2025

GTSinger: A Global Multi-Technique Singing Corpus with Realistic Music Scores for All Singing Tasks

The scarcity of high-quality and multi-task singing datasets significantly hinders the development of diverse controllable and personalized singing tasks, as existing singing datasets suffer from low quality, limited diversity of languages and singers, absence of multi-technique information and realistic music scores, and poor task suitability. To tackle these problems, we present GTSinger, a large global, multi-technique, free-to-use, high-quality singing corpus with realistic music scores, designed for all singing tasks, along with its benchmarks. Particularly, (1) we collect 80.59 hours of high-quality singing voices, forming the largest recorded singing dataset; (2) 20 professional singers across nine widely spoken languages offer diverse timbres and styles; (3) we provide controlled comparison and phoneme-level annotations of six commonly used singing techniques, helping technique modeling and control; (4) GTSinger offers realistic music scores, assisting real-world musical composition; (5) singing voices are accompanied by manual phoneme-to-audio alignments, global style labels, and 16.16 hours of paired speech for various singing tasks. Moreover, to facilitate the use of GTSinger, we conduct four benchmark experiments: technique-controllable singing voice synthesis, technique recognition, style transfer, and speech-to-singing conversion. The demos can be found at http://aaronz345.github.io/GTSingerDemo/. We provide the dataset and the code for processing data and conducting benchmarks at https://huggingface.co/datasets/AaronZ345/GTSinger and https://github.com/AaronZ345/GTSinger.

eess.AS↗

arXiv2025

Language-Codec: Bridging Discrete Codec Representations and Speech Language Models

In recent years, large language models have achieved significant success in generative tasks related to speech, audio, music, and other signal domains. A crucial element of these models is the discrete acoustic codecs, which serve as an intermediate representation replacing the mel-spectrogram. However, there exist several gaps between discrete codecs and downstream speech language models. Specifically, 1) Due to the reconstruction paradigm of the Codec model and the structure of residual vector quantization, the initial channel of the codebooks contains excessive information, making it challenging to directly generate acoustic tokens from weakly supervised signals such as text in downstream tasks. 2) numerous codebooks increases the burden on downstream speech language models. Consequently, leveraging the characteristics of speech language models, we propose Language-Codec. In the Language-Codec, we introduce a Masked Channel Residual Vector Quantization (MCRVQ) mechanism along with improved fourier transform structures and attention blocks, refined discriminator design to address the aforementioned gaps. We compare our method with competing audio compression algorithms and observe significant outperformance across extensive evaluations. Furthermore, we also validate the efficiency of the Language-Codec on downstream speech language models. The source code and pre-trained models can be accessed at https://github.com/jishengpeng/languagecodec .

eess.AS↗