arXiv subjects

Lu Ma

Publications and source records attributed to Lu Ma.

At least 55 records · Page 3Linked to original sources

arXiv2021

Multi-Scale Temporal Convolution Network for Classroom Voice Detection

Teaching with the cooperation of expert teacher and assistant teacher, which is the so-called "double-teachers classroom", i.e., the course is giving by the expert online and presented through projection screen at the classroom, and the teacher at the classroom performs as an assistant for guiding the students in learning, is becoming more prevalent in today's teaching method for K-12 education. For monitoring the teaching quality, a microphone clipped on the assistant's neckline is always used for voice recording, then fed to the downstream tasks of automatic speech recognition (ASR) and neural language processing (NLP). However, besides its voice, there would be some other interfering voices, including the expert's one and the student's one. Here, we propose to extract the assistant' voices from the perspective of sound event detection, i.e., the voices are classified into four categories, namely the expert, the teacher, the mixture of them, and the background. To make frame-level identification, which is important for grabbing sensitive words for the downstream tasks, a multi-scale temporal convolution neural network is constructed with stacked dilated convolutions for considering both local and global properties. These features are concatenated and fed to a classification network constructed by three linear layers. The framework is evaluated on simulated data and real-world recordings, giving considerable performance in terms of precision and recall, compared with some classical classification methods.

cs.SD↗

arXiv2021

Noise Classification Aided Attention-Based Neural Network for Monaural Speech Enhancement

This paper proposes an noise type classification aided attention-based neural network approach for monaural speech enhancement. The network is constructed based on a previous work by introducing a noise classification subnetwork into the structure and taking the classification embedding into the attention mechanism for guiding the network to make better feature extraction. Specifically, to make the network an end-to-end way, an audio encoder and decoder constructed by temporal convolution is used to make transformation between waveform and spectrogram. Additionally, our model is composed of two long short term memory (LSTM) based encoders, two attention mechanism, a noise classifier and a speech mask generator. Experiments show that, compared with OM-LSA and the previous work, the proposed noise classification aided attention-based approach can achieve better performance in terms of speech quality (PESQ). More promisingly, our approach has better generalization ability to unseen noise conditions.

cs.SD↗

arXiv2021

Multi-Scale Attention Neural Network for Acoustic Echo Cancellation

Acoustic Echo Cancellation (AEC) plays a key role in speech interaction by suppressing the echo received at microphone introduced by acoustic reverberations from loudspeakers. Since the performance of linear adaptive filter (AF) would degrade severely due to nonlinear distortions, background noises, and microphone clipping in real scenarios, deep learning has been employed for AEC for its good nonlinear modelling ability. In this paper, we constructed an end-to-end multi-scale attention neural network for AEC. Temporal convolution is first used to transform waveform into spectrogram. The spectrograms of the far-end reference and the near-end mixture are concatenated, and fed to a temporal convolution network (TCN) with stacked dilated convolution layers. Attention mechanism is performed among these representations from different layers to adaptively extract relevant features by referring to the previous hidden state in the encoder long short-term memory (LSTM) unit. The representations are weighted averaged and fed to the encoder LSTM for the near-end speech estimation. Experiments show the superiority of our method in terms of the echo return loss enhancement (ERLE) for single-talk periods and the perceptual evaluation of speech quality (PESQ) score for double-talk periods in background noise and nonlinear distortion scenarios.

cs.SD↗

arXiv2021

Defect Engineered Room-Temperature Ferromagnetism in Quasi-Two-Dimensional Nitrided CoTa2O6

Thermal ammonolysis of quasi-two-dimensional (quasi-2D) CoTa2O6 yields the O2-/N3- and anionic vacancy ordered Co2+Ta5+2O6-xN2x/3$\Box$x/3 (x $\leq$ 0.15) that exhibits a transition from antiferromagnetism to defect engineered above room-temperature ferromagnetism as evidenced by diffraction, spectroscopic and magnetic characterizations. First-principles calculations reveal the origin of ferromagnetism is a particular CoON configuration with N located at Wyckoff position 8j, which breaks mirror symmetry about ab plane. A pressure-induced electronic phase transition is also predicted at around 24.5 GPa, accompanied by insulator-to-metal transition and magnetic moment vanishing.

cond-mat.mtrl-sci↗

arXiv2020

The Wide-field Photometric System of the Nanshan One-meter Telescope

The Nanshan One-meter Wide-field Telescope (NOWT) is a prime focus system located at Nanshan Station of Xinjiang Astronomical Observatories (XAO). The field of view(FOV) was designed to 1.5 degree *1.5 degree, and Johnson-Cousins UBVRI system was chosen as the main Filter set. The telescope has been providing observation services for astronomers since Sept. 2013. Variable source searching and time-domain surveys are the main scientific goals. The system's test results are reported including linearity, dark current, bias, readout noise and gain of the CCD camera. The accurate instrumental calibration coefficients in UBVRI bands was driven with Landolt standard stars during photometric nights. Finally, the limiting magnitudes are given with signal-to-noise ratios and various exposure times for observers.

astro-ph.IM↗

arXiv2020

Spatially-resolved Stellar Population Properties of the M 51-NGC 5195 System from Multi-wavelength Photometric Data

Using multi-band photometric images of M 51 and its companion NGC 5195 from ultraviolet to optical and infrared, we investigate spatially resolved stellar population properties of this interacting system with stellar population synthesis models. The observed IRX is used to constrain dust extinction. Stellar mass is also inferred from the model fitting. By fitting observed spectral energy distributions (SEDs) with synthetical ones, we derive two-dimensional distributions of stellar age, metallicity, dust extinction, and stellar mass. In M 51, two grand-designed spiral arms extending from the bulge show young age, rich metallicity, and abundant dust. The inter-arm regions are filled with older, metal-poorer, and less dusty stellar populations. Except for the spiral arm extending from M 51 into NGC 5195, the stellar population properties of NGC 5195 are quite featureless. NGC 5195 is much older than M 51, and its core is very dusty with $A_V$ up to 1.67 mag and dense in stellar mass surface density. The close encounters might drive the dust in the spiral arm of M51 into the center of NGC 5195.

astro-ph.GA↗

arXiv2020

Physical Properties of H II Regions in M51 from Spectroscopic Observations

M51 and NGC 5195 is an interacting system that can be explored in great details with ground-based telescopes. The H II regions in M51 were observed using the 2.16 m telescope of the National Astronomical Observatories of the Chinese Academy of Sciences and the 6.5 m Multiple Mirror Telescope with spatial resolution of less than $\sim100$ pc. We obtain a total of 113 spectra across the galaxy and combine the literature data of Croxall et al. to derive a series of physical properties, including the gas-phase extinction, stellar population age, star formation rate (SFR) surface density, and oxygen abundance. The spatial distributions and radial profiles of these properties are investigated in order to study the characteristics of M51 and the clues to the formation and evolution of this galaxy. M51 presents a mild radial extinction gradient. The lower gas-phase extinction in the north spiral arms compared to the south arms are possibly caused by the past encounters with the companion galaxy of NGC 5195. A number of H II regions have the stellar age between 50 and 500 Myr, consistent with the recent interaction history by simulations in the literatures. The SFR surface density presents a mild radial gradient, which is ubiquitous in spiral galaxies. There is a negative metallicity gradient of $-0.08$ dex $R_{e}^{-1}$ in the disk region, which is also commonly found in many spiral galaxies. It is supported by the "inside-out" scenario of galaxy formation. We find a positive abundance gradient of 0.26 dex $R_{e}^{-1}$ in the inner region. There are possible reasons causing the positive gradient, including the freezing of the chemical enrichment due to the star-forming quenching in the bulge and the gas infall and dilution due to the pseudobulge growth and/or galactic interaction.

astro-ph.GA↗

arXiv2020

A magneto-optical trap with millimeter ball lenses

We present a magneto-optical trap (MOT) design based on millimeter ball lenses, contained within a metal cube of 0.75$^{\prime \prime}$ side length. We present evidence of trapping approximately $4.2\times 10^5$ of $^{85}$Rb atoms with a number density of $3.2\times 10^9$ atoms/cm$^{3}$ and a loading time of 1.3 s. Measurement and a kinetic laser-cooling model are used to characterize the atom trap design. The design provides several advantages over other types of MOTs: the laser power requirement is low, the small lens and cube sizes allow for miniaturization of MOT applications, and the lack of large-diameter optical beam pathways prevents external blackbody radiation from entering the trapping region.

physics.atom-ph↗

arXiv2020

Acoustic Echo Cancellation by Combining Adaptive Digital Filter and Recurrent Neural Network

Acoustic Echo Cancellation (AEC) plays a key role in voice interaction. Due to the explicit mathematical principle and intelligent nature to accommodate conditions, adaptive filters with different types of implementations are always used for AEC, giving considerable performance. However, there would be some kinds of residual echo in the results, including linear residue introduced by mismatching between estimation and the reality and non-linear residue mostly caused by non-linear components on the audio devices. The linear residue can be reduced with elaborate structure and methods, leaving the non-linear residue intractable for suppression. Though, some non-linear processing methods have already be raised, they are complicated and inefficient for suppression, and would bring damage to the speech audio. In this paper, a fusion scheme by combining adaptive filter and neural network is proposed for AEC. The echo could be reduced in a large scale by adaptive filtering, resulting in little residual echo. Though it is much smaller than speech audio, it could also be perceived by human ear and would make communication annoy. The neural network is elaborately designed and trained for suppressing such residual echo. Experiments compared with prevailing methods are conducted, validating the effectiveness and superiority of the proposed combination scheme.

cs.SD↗

arXiv2020

A Lite Microphone Array Beamforming Scheme with Maximum Signal-to-Noise Ratio Filter

Since space-domain information can be utilized, microphone array beamforming is often used to enhance the quality of the speech by suppressing directional disturbance. However, with the increasing number of microphone, the complexity would be increased. In this paper, a concise beamforming scheme using Maximum Signal-to-Noise Ratio (SNR) filter is proposed to reduce the beamforming complexity. The maximum SNR filter is implemented by using the estimated direction-of-arrival (DOA) of the speech source localization (SSL) and the solving method of independent vector analysis (IVA). Our experiments show that when compared with other widely-used algorithms, the proposed algorithm obtain higher gain of signal-to-interference and noise ratio (SINR).

cs.SD↗

arXiv2020

Competitive Wakeup Scheme for Distributed Devices

Wakeup is the primary function in voice interaction which is the mainstream scheme in man-machine interaction (HMI) applications for smart home. All devices will response if the same wake-up word is used for all devices. This will bring chaos and reduce user quality of experience (QoE). The only way to solve this problem is to make all the devices in the same wireless local area network (WLAN) competing to wake-up based on the same scoring rule. The one closest to the user would be selected for response. To this end, a competitive wakeup scheme is proposed in this paper with elaborately designed calibration method for receiving energy of microphones. Moreover, the user orientation is assisted to determine the optimal device. Experiments reveal the feasibility and validity of this scheme.

cs.SD↗

arXiv2020

Voice Activity Detection Scheme by Combining DNN Model with GMM Model

Due to the superior modeling ability of deep neural network (DNN), it is widely used in voice activity detection (VAD). However, the performance may degrade if no sufficient data especially for practical data could be used for training, thus, leading to inferior ability of adaption to environment. Moreover, large model structure could not always be used in practical, especially for low cost devices where restricted hardware is used. This is on the contrary for Gaussian mixture model (GMM) where model parameters can be updated in real-time, but, with low modeling ability. In this paper, deeply integrated scheme combining these two models are proposed to improve adaptability and modeling ability. This is done by directly combining the results of models and feeding it back, together with the result of the DNN model, to update the GMM model. Besides, a control scheme is elaborately designed to detect the endpoints of speech. The superior performance by employing this scheme is validated through experiments in practical, which give an insight into the advantage of combining supervised learning and unsupervised learning.

cs.SD↗

arXiv2020

Site-testing at Muztagh-ata site I: Ground Meteorology and Sky Brightness

Site-testing is crucial for achieving the goal of scientific research and analysis of meteorological and optical observing conditions is one of the basic tasks of it. As one of three potential sites to host 12-meter Large Optical/infrared Telescope (LOT), Muztagh-ata site which is located on the Pamirs Plateau in west China's Xinjiang began its site-testing task in the spring of 2017. In this paper, we firstly start with an introduction to the site and then present a statistical analysis of the ground-level meteorological properties such as air temperature, barometric pressure, relative humidity, wind speed and direction, recorded by automatic weather station with standard meteorological sensors for two-year long. We also show the monitoring results of sky brightness during this period.

astro-ph.IM↗

arXiv2020

Site-testing at Muztagh-ata site II: Seeing statistics

In this article, we present a detailed analysis of the statistical properties of seeing for the Muztagh-ata site which is the candidate site for hosting future Chinese Large Optical/infrared Telescope (LOT) project. The measurement was obtained with Differential Image Motion Monitor (DIMM) from April 2017 to November 2018 at different heights during different periods. The median seeing at 11 meters and 6 meters are very close but different significantly from that on the ground. We mainly analyzed the seeing at 11 meters monthly and hourly, having found that the best season for observing was from late autumn to early winter and seeing tended to improve during the night only in autumn. The analysis of the dependence on temperature inversion, wind speed, direction also was made and the best meteorological conditions for seeing is given.

astro-ph.IM↗

arXiv2020

Site testing campaign for the Large Optical/infrared Telescope of China: Overview

The Large Optical/infrared Telescope (LOT) is a ground-based 12m diameter optical/infrared telescope which is proposed to be built in the western part of China in the next decade. Based on satellite remote sensing data, along with geographical, logistical and political considerations, three candidate sites were chosen for ground-based astronomical performance monitoring. These sites include: Ali in Tibet, Daocheng in Sichuan, and Muztagh Ata in Xinjiang. Up until now, all three sites have continuously collected data for two years. In this paper, we will introduce this site testing campaign, and present its monitoring results obtained during the period between March 2017 and March 2019.

astro-ph.IM↗

arXiv2019

Qatar Exoplanet Survey: Qatar-8b, 9b and 10b --- A Hot Saturn and Two Hot Jupiters

In this paper we present three new extrasolar planets from the Qatar Exoplanet Survey (QES). Qatar-8b is a hot Saturn, with Mpl = 0.37 Mjup and Rpl = 1.3 Rjup, orbiting a solar-like star every Porb = 3.7 days. Qatar-9b is a hot Jupiter with a mass of Mpl = 1.2 Mjup and a radius of Rpl = 1 Rjup, in a Porb = 1.5 days orbit around a low mass, Mstar = 0.7 Msun, mid-K main-sequence star. Finally, Qatar-10b is a hot, Teq ~ 2000 K, sub-Jupiter mass planet, Mpl = 0.7 Mjup, with a radius of Rpl = 1.54 Rjup and an orbital period of Porb = 1.6 days, placing it on the edge of the sub-Jupiter desert.

astro-ph.EP↗

arXiv2018

Qatar Exoplanet Survey: Qatar-7b -- A Very Hot Jupiter Orbiting a Metal Rich F-Star

We present the discovery of Qatar-7b --- a very hot and inflated giant gas planet orbiting close its parent star. The host star is a relatively massive main sequence F-star with mass and radius Mstar = 1.41 +/- 0.03 Msun and Rstar = 1.56 +/- 0.02 Rsun, respectively, at a distance d = 726 +/- 26 pc, and an estimated age ~1 Gyr. With its orbital period of P = 2.032 days the planet is located less than 5 stellar radii from its host star and is heated to a high temperature Teq ~ 2100 K. From a global solution to the available photometric and radial velocity observations, we calculate the mass and radius of the planet to be Mpl = 1.88 +/- 0.25 Mjup and Rpl = 1.70 +/- 0.03 Rjup, respectively. The planet radius and equilibrium temperature put Qatar-7b in the top 6% of the hottest and largest known exoplanets. With its large radius and high temperature Qatar-7b is a valuable addition to the short list of targets that offer the best opportunity for studying their atmospheres through transmission spectroscopy.

astro-ph.EP↗

arXiv2018

A study of variable stars in the open cluster NGC 1582 and its surrounding field

This paper presents the Charge-Coupled Device time-series photometric observations of the open cluster NGC 1582 and its surrounding field with Johnson B, V and R filters by using Nanshan 1m telescope of Xinjiang Astronomical Observatory. 19 variable stars and 3 variable candidates were detected in a 45 {\times} 48.75 field around the cluster. 12 of the variable stars are newly-discovered variable objects. The physical properties, classifications, and membership of the 22 objects are studied through their light curves, their positions on the color-magnitude diagram, and with the archival data from the Naval Observatory Merged Astrometric Dataset. Among these objects, 5 are eclipsing binary systems, 6 are pulsating variable stars including one known Delta Scuti star and one newly-discovered RR Lyrae star. The distance of the RR Lyrae star is estimated to be 7.9 {\pm} 0.3 kpc, indicates that the star locates far behind the cluster. 4 variable stars are probable members of the cluster, and 13 of the 22 objects are confirmed to be field stars.

astro-ph.SR↗