arXiv subjects

Wei Rao

Publications and source records attributed to Wei Rao.

43 records · Page 3Linked to original sources

arXiv2019

Optimization of Speaker Extraction Neural Network with Magnitude and Temporal Spectrum Approximation Loss

The SpeakerBeam-FE (SBF) method is proposed for speaker extraction. It attempts to overcome the problem of unknown number of speakers in an audio recording during source separation. The mask approximation loss of SBF is sub-optimal, which doesn't calculate direct signal reconstruction error and consider the speech context. To address these problems, this paper proposes a magnitude and temporal spectrum approximation loss to estimate a phase sensitive mask for the target speaker with the speaker characteristics. Moreover, this paper explores a concatenation framework instead of the context adaptive deep neural network in the SBF method to encode a speaker embedding into the mask estimation network. Experimental results under open evaluation condition show that the proposed method achieves 70.4% and 17.7% relative improvement over the SBF baseline on signal-to-distortion ratio (SDR) and perceptual evaluation of speech quality (PESQ), respectively. A further analysis demonstrates 69.1% and 72.3% relative SDR improvements obtained by the proposed method for different and same gender mixtures.

eess.AS↗

arXiv2019

Target Speaker Extraction for Overlapped Multi-Talker Speaker Verification

The performance of speaker verification degrades significantly when the test speech is corrupted by interference speakers. Speaker diarization does well to separate speakers if the speakers are temporally overlapped. However, if multi-talkers speak at the same time, we need the technique to separate the speech in the spectral domain. This paper proposes an overlapped multi-talker speaker verification framework by using target speaker extraction methods. Specifically, given the target speaker information, the target speaker's speech is firstly extracted from the overlapped multi-talker speech by a target speaker extraction module. Then, the extracted speech is passed to the speaker verification system. Experimental results show that the proposed approach significantly improves the performance of overlapped multi-talker speaker verification and achieves 65.7% relative EER reduction.

eess.AS↗

arXiv2017

Gas Eruption Phenomenon Happening from Ga-In Alloy in Electrolyte

We report a gas eruption phenomenon caused by electrolysis of liquid Ga-In alloy in an electrolyte, especially NaOH solution. A volcanic eruption-like blowout of gas occurred from the orifice on the alloy surface. In addition to gas plume, large gas bubbles were also generated and the total gas yield increased as In ratio was increased. It is found that destructiveness of the passivation layer on the Ga-In alloy is critical to gas generation. The mechanism of gas eruption can be ascribed to a galvanic interaction happens owing to passivation film and alloy with different activity connected as electrode in electrolyte. Further investigation demonstrated that the lattice of the film expands because of the incorporation of indium, which brings about the decrease in band gap and finally enhances more gas generation. These findings regain the basic understanding of room temperature liquid metal inside electrolyte.

physics.chem-ph↗

arXiv2017

Transformable Biomimetic Liquid Metal Chameleon

Liquid metal (LM) is of current core interest for a wide variety of newly emerging areas. However, the functional materials thus made so far by LM only could display a single silver-white appearance. Here in this study, the new conceptual colorful LM marbles working like transformable biomimetic chameleons were proposed and fabricated from LM droplets through encasing them with fluorescent nano-particles. We demonstrated that this unique LM marble can be manipulated into various stable magnificent appearances as one desires. And it can also splitt and merge among different colors. Such multifunctional LM chameleon is capable of responding to the outside electric-stimulus and realizing shape transformation and discoloration behaviors as well. Further more, the electric-stimuli has been disclosed to be an easy going way to trigger the release of nano/micro-particles from the LM. The present fluorescent biomimetic liquid metal chameleon is expected to offer important opportunities for diverse unconventional applications, especially in a wide variety of functional smart material and color changeable soft robot areas.

cond-mat.mtrl-sci↗

arXiv2016

Fantastic 4 system for NIST 2015 Language Recognition Evaluation

This article describes the systems jointly submitted by Institute for Infocomm (I$^2$R), the Laboratoire d'Informatique de l'Université du Maine (LIUM), Nanyang Technology University (NTU) and the University of Eastern Finland (UEF) for 2015 NIST Language Recognition Evaluation (LRE). The submitted system is a fusion of nine sub-systems based on i-vectors extracted from different types of features. Given the i-vectors, several classifiers are adopted for the language detection task including support vector machines (SVM), multi-class logistic regression (MCLR), Probabilistic Linear Discriminant Analysis (PLDA) and Deep Neural Networks (DNN).

cs.CL↗

arXiv2013

The Ginger-shaped Asteroid 4179 Toutatis: New Observations from a Successful Flyby of Chang'e-2

On 13 December 2012, Chang'e-2 conducted a successful flyby of the near-Earth asteroid 4179 Toutatis at a closest distance of 770 $\pm$ 120 meters from the asteroid's surface. The highest-resolution image, with a resolution of better than 3 meters, reveals new discoveries on the asteroid, e.g., a giant basin at the big end, a sharply perpendicular silhouette near the neck region, and direct evidence of boulders and regolith, which suggests that Toutatis may bear a rubble-pile structure. Toutatis' maximum physical length and width are (4.75 $\times$ 1.95 km) $\pm$10$\%$, respectively, and the direction of the +$z$ axis is estimated to be (250$\pm$5$^\circ$, 63$\pm$5$^\circ$) with respect to the J2000 ecliptic coordinate system. The bifurcated configuration is indicative of a contact binary origin for Toutatis, which is composed of two lobes (head and body). Chang'e-2 observations have significantly improved our understanding of the characteristics, formation, and evolution of asteroids in general.

astro-ph.EP↗

arXiv2012

ISAR Image Formation Using Sequential Minimization of L0 and L2 Norms

A sparsity-driven algorithm of inverse synthetic aperture radar (ISAR) imaging is proposed. Based on the parametric sparse representation of the received ISAR signal, the problem of ISAR image formation is converted into the joint estimation of the target rotation rate and the sparse power distribution in the spatial domain. This goal is achieved by sequential minimization of L0 and L2 norms, which ensure the sparsest ISAR image and the minimum recovery error, respectively.

cs.IT↗