Search arXiv⌕ Search

arXiv subjects

Ya Li

Publications and source records attributed to Ya Li.

At least 19 recordsLinked to original sources
arXiv2026

From Black Box to Controller: Inference-Time Steering of LLM via Feature Amplification

Inference-time steering is crucial for deploying Large Language Models (LLMs) in real-world applications, where they must flexibly adjust behavior preferences. For instance, a model should dynamically adjust its caution level based on the risk of the scenario. This requirement introduces a challenge: modulating the intensity of LLM preferences. Prompt engineering can require task-specific prompt design, while fine-tuning requires additional optimization and data. These requirements can make repeated behavioral adjustments costly. To address this, we introduce NeuroSteer, a plug-and-play framework that enables flexible, quantitative adjustment of LLM preferences at inference time. NeuroSteer employs a Sparse Autoencoder (SAE) as its controller. Comparing neuron activation frequencies between positive and negative examples isolates "feature neurons" that correspond to target preferences such as politeness and positive sentiment. During inference, the activations of selected SAE features are amplified in the latent representations to steer the output distribution. Users can adjust the strength of this steering on a per-scenario or per-user basis, granting the LLM fine-grained, quantifiable trait modulation. Our experiments show that NeuroSteer can alter sentiment, toxicity, stance, and politeness at inference time. It achieves higher target-direction steering scores than the compared baselines in several settings, with task- and strength-dependent trade-offs in generation quality.

cs.AI↗
arXiv2026

Three-axis Active Magnetic Compensation with Spatial Feedforward Matrix Decoupling and Hysteresis PI Regulation for Suppressing Zeeman Decoherence of Cold Atom Quantum Memories

Cold-atom quantum memories require a near-zero magnetic environment (residual field <1 microtesla) to mitigate Zeeman decoherence. Conventional three-axis active compensation suffers from inter-coil crosstalk and geomagnetic drift, which degrade the uniformity and long-term stability of the zero-field region. To address this, we propose a spatiotemporally cooperative feedback-controlled suppression architecture for magnetic interference. An integrally printed three-dimensional orthogonal coil frame mechanically suppresses inter-axis flux leakage. A spatial feedforward inverse matrix is introduced to algorithmically eliminate first-order cross-coupling. A dual-threshold hysteresis incremental-PI controller filters magnetometer quantization fluctuations and avoids frequent small jitter adjustments of coil driving currents. Hardware ablation experiments, long-duration stability tests and spatial magnetic-field mapping were conducted. Experimental results show that during a 714.4 s continuous steady-hold stage, the mean and maximum residual fields at the central point reach 0.112 microtesla and 0.224 microtesla, respectively. The overlapping Allan deviations of all three components remain below 5 nT for averaging times of 1.9-178.2 s. Based on differential-ferrofluid lattice-scan measurements combined with centre-constrained quadratic vector reconstruction and soft Maxwell regularization, a weak-field spatial distribution with an RMS variation of 5.65 microtesla is fitted within a spherical region of interest of radius 5 mm, where 67.5% of the model volume meets the sub-5 microtesla coverage criterion. For a prolate ellipsoidal region matching the atomic-cloud distribution with full axis lengths of 10, 5, and 5 mm, the upper field bound within 93.7% of the volume is 2.68 microtesla.

quant-ph↗
arXiv2026

VoTSpeech: Decoupling Voice Design from Speech Generation in Instruction-Following TTS

Instruction-following text-to-speech (TTS) typically leaves voice design implicit, as conventional speaker embeddings encode reference audio rather than text-driven targets. We propose VoTSpeech, which decouples voice design from speech generation via an explicit intermediate voice latent. Built on a shared language model backbone, VoTSpeech first predicts a continuous voice latent from text instructions using a flow-based module. This latent serves as a shared condition for both the language model and the Acoustic DiT, guiding autoregressive sequence modeling and acoustic latent generation, respectively, before an audio VAE decoder converts the acoustic latents into waveforms. To capture speaker identity and complementary acoustic nuances, the voice latent is jointly supervised by speaker embeddings and temporally pooled audio VAE features. With task-specific adaptation, VoTSpeech performs strongly on InstructTTSEval-ZH, demonstrating competitive capabilities against strong open-source models while maintaining high naturalness. These results confirm that explicit voice latent design and its conditioning interface effectively empower text-driven voice design for controllable speech synthesis. Code, model, and audio samples are available at https://github.com/ywbn/VoTSpeech .

cs.SD↗
arXiv2026

SemiCodec: Bridging Speech Tokenization and High-fidelity Reconstruction with Semi-discrete Representations

Speech representation learning requires a discrete modeling interface while preserving high waveform fidelity. Existing approaches rarely achieve both: discrete tokens ease modeling but lose fine-grained acoustic information, while continuous latents preserve fidelity but require specialized generative objectives. We introduce SemiCodec, a novel Semi-discrete neural speech Codec with synchronized discrete tokens and continuous latents. Direct joint optimization may create a reconstruction shortcut through the continuous pathway, leaving the discrete tokens insufficiently grounded. To address this issue, we propose a Preserve-then-Ground training paradigm that first establishes a high-fidelity synthesis space and then grounds SSL-guided discrete tokens within it. Experiments show that SemiCodec outperforms fully discrete codecs in reconstruction quality and intelligibility while remaining competitive with continuous representations in acoustic fidelity and speaker preservation. Audio samples are available at https://pennyw1-se.github.io/SemiCodec-Demo-Page/

eess.AS↗
arXiv2026

DPPM: Dual-Path Parametric Memory for Personalized Language Models

Long-term personalization requires language models to use interaction history to track users' preferences across sessions. Parametric memory encodes this interaction history into model parameters or adapters, reducing the need to include it in the inference context. However, independent context compilation leaves cross-session integration unspecified, while recurrent updates can attenuate earlier evidence. To address these challenges, we propose Dual-Path Parametric Memory (DPPM). Its Evidence path directly pools representations of the interaction history to preserve earlier evidence, while its Delta path sequentially updates an associative state to capture changes. Fusing both outputs produces history-conditioned LoRA adapters that combine evidence accumulation with ordered revision. Across multiple backbones, DPPM outperforms the evaluated baselines, achieving 54.22% on PersonaMem-v2 and 86.79% on PrefEval. These results suggest that DPPM provides a simple and effective design choice for cross-session personalized parametric memory.

cs.CL↗
arXiv2026

Measures of Reliability Risk for the Australian Energy Sector

The article identifies the desired properties of the reliability risk metric for the Australian Energy Sector. It proposes a comprehensive set of metrics designed to measure various aspects of the reliability risk when a large portion of power generation is composed of variable renewable energy (VRE). The suggested methods aim to balance the tradeoff between economic benefit and risk management, effectively address the tail risk, and interpret the severity of the outage in a direct way. Properties of the considered reliability metrics are investigated by using simulation studies that use time series of expected unserved energy values. Sample distributions, dependency structures and other statistical properties of the metrics are studied. The results suggest new tools and approaches that can be used to increase the efficiency of the Australian Energy Sector.

stat.ME↗
arXiv2026

From Reliable Text to Real Voices: Trust-Aware Progressive Adaptation for Low-Resource TTS

Low-resource text-to-speech (TTS) adaptation is constrained by scarce paired data and costly manual transcription. Existing fixed-voice TTS systems can provide relatively accurate pronunciation, but their synthetic speech offers limited speaker diversity and may exhibit flat prosody. Real recordings provide natural prosody and diverse voices, yet their automatic speech recognition (ASR) pseudo-labels may contain transcription errors. We find that supervision order affects content accuracy and speaker similarity. We propose trust-aware progressive adaptation: synthetic-to-real adaptation first establishes text-speech correspondences, then restores reference-speaker control using real speech. Transcript-agreement weighting uses agreement between two fixed ASR systems as a proxy for pseudo-label reliability to limit noisy supervision. Experiments with FireRedTTS3 on Burmese and Lao and OmniVoice on Burmese show improved content accuracy with high naturalness and competitive speaker similarity. Jointly considering supervision order and pseudo-label reliability when combining synthetic and real speech offers a practical path to zero-shot voice cloning in low-resource languages with less manual transcription. Audio demos are available at https://insiderx-pro.github.io/S2R-Adaptation-TTS/

cs.SD↗
arXiv2026

Stabilizing Instruction Supervision for Instruct-TTS via Controllable Diversification and Drift Filtering

Instruct-TTS systems expand structured style labels into natural-language training instructions through LLM rewriting, yet we find that over 40% of unconstrained rewrites contain semantic drift that corrupts supervision and weakens generalization. We formalize this problem as instruction supervision instability and propose a data-centric stabilization recipe that jointly improves coverage and fidelity through three mechanisms: controllable instruction diversification for systematic expansion, LLM-based drift filtering for quality control, and attribute-aligned supervision that grounds prosody control in acoustic perturbations. On the Chinese split of InstructTTSEval, our recipe raises instruction-following from 34.5% without fine-tuning and 51.0% with naive fine-tuning to 56.4%, while constrained rewriting reduces drift from 40.4% to 15.4%. Ablations confirm the three mechanisms are complementary, and the drift taxonomy may generalize to instruction-driven generation beyond TTS.

cs.SD↗
arXiv2026

CLASVS: Continuous-Latent Autoregression for Melody-Preserving Lyric Editing in Singing Voice Synthesis

Reference-conditioned melody-preserving lyric editing replaces words while retaining a performance's timing, singer identity, and naturalness. Continuous-latent autoregression avoids finite codebooks and offers stepwise generation with learned stopping. Editing creates a conflict absent from ordinary reconstruction: training pairs reference cues with original lyrics, whereas inference asks revised lyrics to override source-lyric-correlated cues; one source-following patch can propagate through AR history. We introduce CLASVS. Its State-Control-Transition (SCT) routing keeps target-lyric and reference-melody controls persistent, returns semantic feedback on phonetic progress to the causal planner, and confines the previous latent patch to the local Transition. Progressive State-Control Grounding (PSCG) learns this contract through paired-edit-free, content-consistent Mandarin reconstruction. On two Mandarin benchmarks, CLASVS improves all four operations over discrete-AR Vevo2 and reduces macro-PER by 46.2%, while maintaining melody, singer similarity, and perceptual quality. Together, these results establish a strong continuous-AR operating point for score-annotation-free lyric edits and a basis for broader stepwise control. Audio demonstrations are available on our project page: https://piedpiperg.github.io/clasvs-demo/.

cs.SD↗
arXiv2026

SceneTTS-Bench: A Benchmark for Scene-Level TTS in Drama Dubbing

Text-to-speech systems are increasingly used for drama dubbing, yet evaluation protocols remain sentence-level, leaving critical scene-level behaviors insufficiently measured. We present SceneTTS-Bench, a benchmark that evaluates TTS along three dimensions: timbre consistency across character turns, emotional expressiveness on high-tension utterances, and rhythm coherence under segmented long-form synthesis. The corpus comprises real-world and generated drama scripts totaling 160 bilingual scenes and approximately 10,300 utterances, with real-world scripts serving as the primary source (100 scenes) and generated scripts as a supplementary source (60 scenes), demonstrating the framework's extensibility through synthetic data augmentation. A backend-agnostic Canonical Intermediate Representation ensures fair cross-system comparison. Three automatic pipelines produce per-utterance diagnostics: Speaker Consistency Score for timbre-drift detection, Under-Acting Ratio for under-acting identification, and Rate Discontinuity Ratio for rate-discontinuity quantification. Experiments on four TTS systems confirm that each system exhibits distinct weaknesses and that scene-level rankings diverge substantially from sentence-level metrics. Benchmark resources are publicly available at https://piedpiperg.github.io/scenetts-bench/ .

cs.SD↗
arXiv2026

Large CP violation in $Λ_b\rightarrow ΛD$ decays and extraction of the Cabibbo-Kobayashi-Maskawa angle $γ$

Motivated by the first observation of CP violation in $b$-baryon decays, the search for baryonic decays exhibiting large CP violation will be a primary focus in the coming years. We propose that significant CP-violating effects exist in the decay $Λ_b \to ΛD$, where $D$ denotes a CP eigenstate of the $D^0 - \bar{D}^0$ system. The predicted CP asymmetries for both the CP-even and CP-odd modes can reach magnitudes as large as $50\%$, making these decays promising targets for measurement at the LHCb experiment. Additionally, we predict for the first time several nonzero CP-violating observables associated with angular distribution parameters, providing valuable complementary information in the search for CP violation in baryon decays. Furthermore, we propose a novel strategy to extract the CKM angle $γ$ by combining data on angular distribution parameters and decay rates from the relevant channels. We emphasize that $Λ_b \rightarrow ΛD$ decays are among the most promising candidates for determining $γ$ in the baryon sector. Our findings may offer new insights for future theoretical and experimental investigations.

hep-ph↗
arXiv2026

MIDAS: Mutual Information Disentanglement with Uncertainty-Aware Fusion for Incomplete Multimodal Sentiment Analysis

Most existing multimodal sentiment analysis approaches assume access to complete multimodal inputs. However, real-world applications frequently encounter incomplete or corrupted modalities, posing a critical challenge. Although several methods have been proposed to tackle this issue, they mainly rely on data imputation and heuristic coordination constraints, which fail to effectively extract and leverage task-relevant information from the incomplete multimodal data. To address this challenge, we propose a unified framework termed Mutual Information Disentanglement with uncertainty-Aware fuSion (MIDAS), which effectively restructures multimodal representations under incomplete conditions. MIDAS adopts a variational modeling strategy to represent each modality with multivariate Gaussian latent variables and further decomposes them into shared and exclusive factors. To obtain reliable representations, we design a minimax objective that minimizes the mutual information between shared and exclusive spaces for stable disentanglement, while maximizing the mutual information among shared spaces across modalities to enhance semantic alignment. In addition, an uncertainty-aware fusion mechanism is introduced, where posterior variance is leveraged as a reliability indicator to adaptively weight latent features during fusion, ensuring robust integration even when modalities are incomplete. Extensive experiments on three widely used datasets show that MIDAS achieves strong and consistent performance gains over competitive baselines across a wide range of incomplete settings, demonstrating its effectiveness and robustness for incomplete data scenarios.

cs.AI↗
arXiv2026

Any3DAvatar: Fast and High-Quality Full-Head 3D Avatar Reconstruction from Single Portrait Image

Reconstructing a complete 3D head from a single portrait remains challenging because existing methods still face a sharp quality-speed trade-off: high-fidelity pipelines often rely on multi-stage processing and per-subject optimization, while fast feed-forward models struggle with complete geometry and fine appearance details. To bridge this gap, we propose Any3DAvatar, a fast and high-quality method for single-image 3D Gaussian head avatar generation, whose fastest setting reconstructs a full head in under one second while preserving high-fidelity geometry and texture. First, we build AnyHead, a unified data suite that combines identity diversity, dense multi-view supervision, and realistic accessories, filling the main gaps of existing head data in coverage, full-head geometry, and complex appearance. Second, rather than sampling unstructured noise, we initialize from a Plücker-aware structured 3D Gaussian scaffold and perform one-step conditional denoising, formulating full-head reconstruction into a single forward pass while retaining high fidelity. Third, we introduce auxiliary view-conditioned appearance supervision on the same latent tokens alongside 3D Gaussian reconstruction, improving novel-view texture details at zero extra inference cost. Experiments show that Any3DAvatar outperforms prior single-image full-head reconstruction methods in rendering fidelity while remaining substantially faster.

cs.CV↗
arXiv2026

MeloCodec: Harnessing Melodic Priors for High-Fidelity Singing Voice Representation

Neural audio codecs serve as fundamental tokenizers for LLM-based audio generation. While semantic priors are widely exploited to enhance linguistic intelligibility, the integration of explicit acoustic priors remains underexplored, limiting synthesis fidelity in frequency-sensitive domains. To address this gap, we introduce MeloCodec, a novel framework designed to effectively incorporate melodic priors, a critical form of acoustic information for singing. To address the optimization instability typically caused by the direct fusion of such explicit priors, we propose a Tokenize-then-Fuse paradigm that pre-trains a discrete melodic branch to lock in structures before feature fusion. To robustly realize this paradigm, we further propose a two-stage training strategy that prevents codebook collapse and ensures stable convergence. Experiments show that MeloCodec outperforms baselines in singing voice representation, improving pitch consistency and enabling controllable pitch manipulation with minimal timbre degradation.

cs.SD↗
arXiv2026

Polarization fractions and helicity-dependent CP asymmetries in $B_{(s)} \to ρρ, ρK^\ast$ and $K^\ast K^\ast$ decays

In this paper, we present a phenomenological analysis of $B_{(s)} \to ρρ, ρK^\ast$ and $K^\ast K^\ast$ decays using state-of-the-art perturbative QCD (pQCD) calculations. Our study is primarily motivated by recent polarization measurements from the LHCb and Belle II collaborations, which have significantly improved the precision of polarization fractions and enabled the first full determination of polarization-dependent CP asymmetries. This work extends the comprehensive pQCD study of charmless two-body $B$ decays reported in our previous paper [Chin. Phys. C 46 (2022) 123103], with a particular focus on polarization observables, especially the CP asymmetries in each helicity state, which reflect distinct orbital angular momentum configurations between the two vector mesons. Our predictions for the branching ratios and longitudinal polarization fractions in the $B^0 \to K^{\ast 0} {\bar K}^{\ast 0}$ and $B^+ \to ρ^0 K^{\ast +}$ modes are in good agreement with the new experimental data. However, the calculated longitudinal polarization fraction for $B_s \to K^{\ast 0} {\bar K}^{\ast 0}$ is significantly larger than the LHCb measurement. Moreover, the predicted (helicity-dependent) CP asymmetries in $B^+ \to ρ^0 K^{\ast +}$ are about $30 \%$ smaller than the observed values. These discrepancies point to a rich interplay between different topological amplitudes, highlighting the need for further theoretical investigation to resolve the long-standing polarization puzzle in two-body $B$ decays into vector mesons.

hep-ph↗
arXiv2026

SafeRelBench: A Spatial-Relation-Aware Benchmark for Process-Level Safety in VLM-Driven Embodied Agents

Vision-language models (VLMs) are increasingly used as the reasoning backbone of embodied agents, enabling robots to interpret visual scenes, follow language instructions, and plan multi-step actions. In household environments, however, safety depends not only on recognizing objects, but also on how actions change the physical scene over time. Existing embodied safety evaluations largely focus on static risk recognition, unsafe instruction refusal, or final-state task completion. As a result, process-level safety failures induced by spatial relations such as support, containment, and proximity remain insufficiently studied. To address this gap, we introduce SAFERELBENCH, a spatial-relation-aware safety benchmark with 507 executable evaluation samples, including 248 spatial-relation samples and 259 non-spatial control samples. Using SAFERELBENCH to evaluate seven open- and closed-source VLM-driven embodied agents, we find a substantial gap between task success and process-level safety compliance: models often complete the requested task while violating process-level safety constraints. Unlike prior benchmarks, SAFERELBENCH explicitly tests whether agents satisfy safety conditions before risk-prone actions, making spatial relations a core dimension in embodied safety assessment. More broadly, our results show that safe embodied intelligence requires not only stronger perception and planning, but also reliable reasoning about how object relations shape risk during interaction.

cs.RO↗
arXiv2026

Capacity Bounds and High-SNR Characterization for MIMO-OWC Channels Under Average-Power Constraint

This paper investigates the capacity of multipleinput multiple-output (MIMO) optical wireless communication (OWC) channels under a total average-power constraint. Since different nonnegative input vectors can be mapped to the same image vector and thus induce the same output distribution, we formulate a nonnegative basis pursuit (NN-BP) problem to identify the minimum-l1-norm input vector for each image vector. Based on the NN-BP characterization, we derive an equivalent expression for the channel capacity in terms of the image-vector distribution. We then establish computable lower and upper capacity bounds for both nT >= nR and nT < nR cases, and prove that the proposed bounds are asymptotically tight in the high signal-to-noise ratio (SNR) regime. Numerical results for indoor and outdoor OWC scenarios demonstrate that the proposed bounds improve upon existing ones and close the constant gap in the high-SNR regime.

cs.IT↗
arXiv2026

AffectCodec: Emotion-Preserving Neural Speech Codec with Block-Diagonal Residual FSQ

Neural speech codecs have become the discrete interface between raw audio and speech language models, yet they remain optimized primarily for acoustic reconstruction fidelity, which leaves emotion-relevant cues vulnerable to being discarded during quantization, limiting the affective capacity of downstream models. We trace this degradation to two mechanisms: reconstruction-driven bit allocation under limited bitrate and cross-stream leakage in concatenation-based codecs, where acoustic gradients can overwrite nominally emotion-reserved dimensions. We propose AffectCodec, an emotion-preserving neural speech codec built on Block-Diagonal Residual Finite Scalar Quantization (BD-RFSQ). By imposing block-diagonal input and output projections over emotion and acoustic subspaces, BD-RFSQ transforms bit allocation from implicit and loss-driven to explicit and structurally guaranteed, while still preserving a flat token interface for downstream speech language models. AffectCodec further combines this structurally constrained quantizer with multi-granularity emotion conditioning and multi-rate training, enabling robust affect preservation at low bitrates. Experiments across multiple emotional speech benchmarks show that AffectCodec substantially improves emotion preservation, especially in the low-bitrate regime, while maintaining competitive acoustic quality and intelligibility. These results suggest that structurally protected quantization is an effective principle for preserving emotion-relevant information and may provide a general route toward attribute-aware neural speech compression.

cs.SD↗