Xubo Liu

ORCID: 0000-0002-2558-0959
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Magnetic Properties of Alloys
  • Music and Audio Processing
  • Magnetic and transport properties of perovskites and related materials
  • Speech and Audio Processing
  • Rare-earth and actinide compounds
  • Magnetic properties of thin films
  • Speech Recognition and Synthesis
  • Hydrogen Storage and Materials
  • Magnetic Properties and Applications
  • Natural Language Processing Techniques
  • Video Analysis and Summarization
  • Advanced Condensed Matter Physics
  • Metallic Glasses and Amorphous Alloys
  • Subtitles and Audiovisual Media
  • Multimodal Machine Learning Applications
  • Metallurgical and Alloy Processes
  • Music Technology and Sound Studies
  • Shape Memory Alloy Transformations
  • Advanced Adaptive Filtering Techniques
  • Speech and dialogue systems
  • Heusler alloys: electronic and magnetic properties
  • Superconducting Materials and Applications
  • Water Quality Monitoring Technologies
  • Anomaly Detection Techniques and Applications
  • Animal Vocal Communication and Behavior

University of Surrey
2021-2025

Ames National Laboratory
2016-2024

Signal Processing (United States)
2023-2024

Qingdao National Laboratory for Marine Science and Technology
2022-2024

Laoshan Laboratory
2024

Ocean University of China
2022-2024

Hubei Zhongshan Hospital
2024

Wuhan University
2024

Henan Normal University
2024

Shanghai East Hospital
2024

Text-to-audio (TTA) system has recently gained attention for its ability to synthesize general audio based on text descriptions. However, previous studies in TTA have limited generation quality with high computational costs. In this study, we propose AudioLDM, a that is built latent space learn the continuous representations from contrastive language-audio pretraining (CLAP) latents. The pretrained CLAP models enable us train LDMs embedding while providing as condition during sampling. By...

10.48550/arxiv.2301.12503 preprint EN other-oa arXiv (Cornell University) 2023-01-01

Cobalt nanowires with high aspect ratio have been synthesized via a solvothermal chemical process. Based on the shape anisotropy and orientation of nanowire assemblies, record room-temperature coercivity 10.6 kOe has measured in Co diameter about 15 nm mean length 200 nm. As result, energy product wires reaches 44 MGOe. It is discovered that morphology uniformity key to achieving density. Nanowires this type are ideal building blocks for future bonded, consolidated thin film magnets density...

10.1038/srep05345 article EN cc-by-nc-nd Scientific Reports 2014-06-18

Abstract Drug loaded dissolving microneedles (DMNs) fabricated with water soluble polymers have received increasing attentions as a safe and efficient transdermal drug delivery system. Usually, to reach high efficiency, an ideal distribution is gathering more drugs in the tip or top part of DMNs. In this work, we introduce easy new method bubble controlled size into body The introduction bubbles can prevent diffusion whole MNs. heights are well from 75 μm 400 just by changing mass...

10.1038/srep38755 article EN cc-by Scientific Reports 2016-12-08

Audio-text retrieval aims at retrieving a target audio clip or caption from pool of candidates given query in another modality.Solving such cross-modal task is challenging because it not only requires learning robust feature representations for both modalities, but also capturing the fine-grained alignment between these two modalities.Existing models are mostly optimized by metric objectives as them attempt to map data an embedding space, where similar close together and dissimilar far...

10.21437/interspeech.2022-11115 article EN Interspeech 2022 2022-09-16

In this paper, we introduce the task of language-queried audio source separation (LASS), which aims to separate a target from an mixture based on natural language query (e.g., "a man tells joke followed by people laughing"). A unique challenge in LASS is associated with complexity description and its relation sources. To address issue, proposed LASS-Net, end-to-end neural network that learned jointly process acoustic linguistic information, consistent mixture. We evaluate performance our...

10.21437/interspeech.2022-10894 article EN Interspeech 2022 2022-09-16

10.1016/j.jallcom.2023.170258 article EN publisher-specific-oa Journal of Alloys and Compounds 2023-04-24

Neutron-diffraction and magnetization measurements have been carried out on a series of samples the magnetorefrigerant ${\text{Mn}}_{1+y}{\text{Fe}}_{1\ensuremath{-}y}{\text{P}}_{1\ensuremath{-}x}{\text{Ge}}_{x}$. The data reveal that ferromagnetic paramagnetic phases correspond to two very distinct crystal structures, with magnetic-entropy change as function magnetic field or temperature being directly controlled by phase fraction this first-order transition. By tuning physical properties...

10.1103/physrevb.79.014435 article EN Physical Review B 2009-01-26

Doping with Dy/Tb is a current solution to improve the coercivity and operating temperature of NdFeB magnets, by increasing magnetocrystalline anisotropy (Nd, Dy/Tb)2Fe14B (2:14:1) phase. The efficiency doping depends on distribution in multi-phase microstructure magnet. To understand control redistribution, partitioning between 2:14:1 Nd-rich phases has been studied first-principles density functional calculation. total energy calculations indicate that Dy Tb prefer enter 4f sites...

10.1063/1.3670054 article EN Journal of Applied Physics 2012-02-06

We report magnetic properties and microstructure of high energy-product MnBi bulk magnets fabricated by low-temperature ball-milling warm compaction technique. A maximum energy product (BH)max 8.4 MGOe a coercivity 6.2 kOe were obtained in the magnet at room temperature. Magnetic characterization elevated temperatures showed an increase to 16.2 while value decreased 6.8 400 K. Microstructure revealed that consist oriented uniform nanoscale grains with average size about 50 nm.

10.1063/1.4942955 article EN cc-by AIP Advances 2016-02-23

Automated audio captioning is a cross-modal translation task that aims to generate natural language descriptions for given clips. This has received increasing attention with the release of freely available datasets in recent years. The problem been addressed predominantly deep learning techniques. Numerous approaches have proposed, such as investigating different neural network architectures, exploiting auxiliary information keywords or sentence guide caption generation, and employing...

10.1186/s13636-022-00259-2 article EN cc-by EURASIP Journal on Audio Speech and Music Processing 2022-10-09

Doping a small amount of Al can effectively enhance coercivity in Nd-Fe-B magnets. We investigated the partitioning behavior and its effect on using first principles DFT (density functional theory) calculation micromagnetic simulation. The calculated substitution energies Fe by are negative at crystallographic sites 4c 8j2 while they positive values other Nd2Fe14B (2:14:1), implying solubility 2:14:1. Further, prefers to segregate grain boundary (GB) stabilize Nd-Fe-Al phase with...

10.1063/9.0000645 article EN cc-by AIP Advances 2024-01-01

In this work, we develop a scalable approach for flexible latent factor model high-dimensional dynamical systems. Each process has its own correlation and variance parameters, the orthogonal loading matrix can be either fixed or estimated. We utilize an that avoids computing inversion of posterior covariance at each time Kalman filter, derive closed-form expressions in expectation-maximization algorithm parameter estimation, which substantially reduces computational complexity without...

10.48550/arxiv.2501.01324 preprint EN arXiv (Cornell University) 2025-01-02

Anomalous sound detection (ASD) encounters difficulties with domain shift, where the sounds of machines in target domains differ significantly from those source due to varying operating conditions. Existing methods typically employ classifiers enhance performance, but they often overlook influence domain-unrelated information. This oversight can hinder model's ability clearly distinguish between domains, thereby weakening its capacity differentiate normal abnormal sounds. In this paper, we...

10.48550/arxiv.2501.01604 preprint EN arXiv (Cornell University) 2025-01-02

Structured droplets, constructed and stabilized by interfacial self-assembly jamming of colloidal particles, have shown potential applications in biphasic reactors towards fabrication functional materials. However, the macroscopic...

10.1039/d5ta00494b article EN Journal of Materials Chemistry A 2025-01-01

Superconductivity beyond the conventional Bardeen-Cooper-Schrieffer (BCS) framework often emerges out of a normal state that is accompanied by exotic magnetism and thereby displays many exceptional transport thermodynamic properties. Here we report heavy $d$-electron superconductor Rh$_{17}$S$_{15}$ characterized weak \textit{ferromagnetism} persists up to room temperature. We show broad hump in its resistivity likely results from Kondo interaction conduction electrons with this novel...

10.48550/arxiv.2502.12416 preprint EN arXiv (Cornell University) 2025-02-17

10.1109/icassp49660.2025.10888266 article EN ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2025-03-12

10.1109/icassp49660.2025.10888719 article EN ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2025-03-12

10.1109/icassp49660.2025.10888065 article EN ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2025-03-12

10.1016/s0304-8853(03)00207-5 article EN Journal of Magnetism and Magnetic Materials 2003-04-23

10.1016/j.jmmm.2004.10.100 article EN Journal of Magnetism and Magnetic Materials 2004-11-13
Coming Soon ...