Abstract:When faced with a mixed audio, people are often interested in the voice of one person instead of all the voices. Speaker extraction is precisely for this situation, extracting the voice of the target speaker in a multi-speaker environment, in order to imitate human selective auditory attention. In previous models, extraction is generally performed in the frequency domain, and the time domain signal is reconstructed according to the extracted amplitude and estimated phase spectrum. But this method will be affected by the phase estimation when reconstructing the time domain signal. With reference to Conv-TasNet and SpEx, this paper uses a speaker extraction network (F3S) in the time domain. The network converts the mixed speech into embedding coefficients instead of decomposing the speech signal into amplitude and phase spectra, thereby avoiding phase estimation problems. On this basis, this paper uses the μ-law compression and expansion algorithm to process the data, aiming to improve the training speed and extraction speed. The experimental results show that the F3S network in this paper can effectively improve the training speed and extraction speed while maintaining a good scale-invariant SDR (SI-SDR).

Alert Name:
Alerting to:
Authentication email will be sent to your email address in 24 hours
Frequency:
Email Message Format:	Plain text Graphical(HTML)

Complete the form below and we will recommend the selected titles to your friends on your behalf. * Indicates a required field.
Your name*:
Your email address*:
Recipient's name*:
Recipient's email address*:
(multiple recipient's names and email addresses should be separated with semicolons)
Your comments:	I thought you would find the page(s) useful.

Your name:
Your email address:
Recipient's name:
Recipient's email address:
(multiple recipient's names and email addresses should be separated with semicolons)
Your comments:	I thought you would find this page useful.

Disclaimer: This message was sent to your friend using the "Send it to a friend" facility on the Sciencepaper Online’ WWW site, http://www.paper.edu.cn/en. The Sciencepaper Online is not responsible for the content of this email, and anything said in this email does not necessarily reflect the Sciencepaper Online's views.


	1. Research on Single-channel Target Speech Extraction Using μ-Law Algorithm
	Linhui Qiu,Jing Wang
	Computer Science and Technology 17 January 2022
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (736K B)

	Abstract:When faced with a mixed audio, people are often interested in the voice of one person instead of all the voices. Speaker extraction is precisely for this situation, extracting the voice of the target speaker in a multi-speaker environment, in order to imitate human selective auditory attention. In previous models, extraction is generally performed in the frequency domain, and the time domain signal is reconstructed according to the extracted amplitude and estimated phase spectrum. But this method will be affected by the phase estimation when reconstructing the time domain signal. With reference to Conv-TasNet and SpEx, this paper uses a speaker extraction network (F3S) in the time domain. The network converts the mixed speech into embedding coefficients instead of decomposing the speech signal into amplitude and phase spectra, thereby avoiding phase estimation problems. On this basis, this paper uses the μ-law compression and expansion algorithm to process the data, aiming to improve the training speed and extraction speed. The experimental results show that the F3S network in this paper can effectively improve the training speed and extraction speed while maintaining a good scale-invariant SDR (SI-SDR).
	TO cite this article:Linhui Qiu,Jing Wang. Research on Single-channel Target Speech Extraction Using μ-Law Algorithm[OL].[17 January 2022] http://en.paper.edu.cn/en_releasepaper/content/4756081


	2. Adaptive speech enhancement based on SNR perception in non-stationary noise scenarios
	Chen Zhishuai,Wang Jing
	Computer Science and Technology 14 January 2022
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (682K B)

	Abstract:Speech enhancement is always a hot topic in the field of speech processing. In real life, interference signals are usually non-stationary or even burst. Therefore, it is of great significance to study speech enhancement technology in non-stationary noise scene to solve practical problems. The speech enhancement algorithm for the following questions, first, is currently on the market more studies on speech enhancement based on the assumption of stationary noise, but in real life, the environment is rapidly changing, noise cannot be ideally unchanged, second, because of the non-stationary noise is hard to predict, in the case of low SNR, it is difficult to maintain the signal distortion after speech enhancement. Aiming at the above problems, this paper proposes an adaptive speech enhancement method based on SNR perception in non-stationary noise scenes to improve the adaptability of speech enhancement in complex non-stationary scenes. At the same time, the neural network is designed to calculate the signal-to-noise ratio and improve the processing capability of the model.
	TO cite this article:Chen Zhishuai,Wang Jing. Adaptive speech enhancement based on SNR perception in non-stationary noise scenarios[OL].[14 January 2022] http://en.paper.edu.cn/en_releasepaper/content/4756093

Saved Papers

Saved Papers


	3. P-SANET: A HIGH-PRECISION REALTIME POINT CLOUD SEMANTIC SEGMENTATION FRAMEWORK
	GOU Xiaofeng,JIAO Jichao,ZHANG Chengkai
	Computer Science and Technology 10 January 2022
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (1440K B)

	Abstract:Perception in autonomous system is an important task to guide decision execu-tion. Lidar point cloud is a type of dataset to complete perception task, it is rich in original information, easy to collect, and convenient to store. Compared to camera image, point cloud contains precise spatial information and adapts to various en-vironments, nevertheless, more information means more computing power con-sumption. The processing speed and accuracy are two key metrics of neural net-work framework. The traditional methods have to pay the price of reducing accu-racy for increasing processing speed. Though some frameworks preprocess point cloud into projected image, the 2D image tensor also contains a large number of redundant channel features in the traditional 2Dconvolution operation. In this pa-per, we propose a point clouds semantic segmentation framework, we replace the standard convolutional layer with a new sub-module, and it greatly reduces the amount of computation, besides, we introduce a sub-module to fuse the coordi-nate values and middle tensors. The framework in this paper is divided into three parts: spherical projection preprocessing module, En-Decoder module and data post-processing module. We use the SemanticKITTI dataset to conduct experi-ments, and the results show that our framework outperforms other frameworks both in prediction accuracy and prediction speed. We also use sparse point cloud dataset to test the generalization of our framework, and the experiments show that it performs better than other frameworks. Code is available at: https://github.com/windtries/P-SANet
	TO cite this article:GOU Xiaofeng,JIAO Jichao,ZHANG Chengkai. P-SANET: A HIGH-PRECISION REALTIME POINT CLOUD SEMANTIC SEGMENTATION FRAMEWORK[OL].[10 January 2022] http://en.paper.edu.cn/en_releasepaper/content/4755947


	4. An Abstract Syntax Tree Construction Analysis Model Based on JavaCC
	WANG Guolong,JIN Dahai
	Computer Science and Technology 09 January 2022
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (1286K B)

	Abstract:Abstract syntax tree (AST) was an important intermediate expression form of code. Aiming at the deficiencies of existing lexical and syntax analysis tool JavaCC (Java Compiler Compiler) that it did not support the new C++ standard and did not have a good error handling mechanism, an abstract syntax tree construction analysis model based on JavaCC was proposed in this paper. First, C++ grammar rules through the EBNF (Extended Backus-Naur Form) are defined, and a rule expansion method is proposed to deal with future syntax updates. Secondly, a syntax tree is constructed based on the JavaCC by generating lexical and grammar parser, and the key information is extracted through traversing the syntax tree. Thirdly, some strategies are proposed such as embedding Java code, skipping mismatched grammatical fragments, and constructing the feature hash of syntax tree nodes for error handling. Finally, a prototype system based on this model is designed and developed in this paper, and the effectiveness of the model in both time and space dimensions is verified through open source engineering experiments adopting C++ new standard.
	TO cite this article:WANG Guolong,JIN Dahai. An Abstract Syntax Tree Construction Analysis Model Based on JavaCC[OL].[ 9 January 2022] http://en.paper.edu.cn/en_releasepaper/content/4756084


	5. SOFT-AlignUNet: A Lightweight Transformer with Feature Alignment
	WU Rui-Jia,ZHANG Hong-Gang
	Computer Science and Technology 04 January 2022
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (930K B)

	Abstract:Transformer, the prevalent backbone architecture in natural language processing, has been adopted in various vision tasks since the proposition of vision transformer. The performance of transformer has been proved to be almost the same as CNN's and even be better with large enough dataset. However, the initial vision transformer suffered from the straightforward structure, which requires large parameters and expensive computation cost, especially in the dense prediction task. This paper concentrates on medical image semantic segmentation task. In medical scene, UNet is always the popular backbone and many researchers proposed transformer-CNN or pure transformer UNet model recently. But the inherent feature misalignment caused by resizing feature maps and concatenation is still lack of focus. In this paper, a lightweight transformer-CNN hybrid UNet, SOFT-AlignUNet (SOFT-AU) , is proposed to solve above issues. On one hand, a novel softmax-free transformer, which reduces the calculation cost to be linear to the patch number, is introduced into UNet architecture to alleviate the computation cost at a large extent. On the other hand, the feature misalignment is taken into consideration and a river-like Feature Alignment Flow is proposed to generate spatial deviation and correct the features. The architecture achieves strongly competitive results on public Synapse and DRIVE dataset with pretty light model size and computation requirement. The results show that this is a pretty promising network for future deployment in reality.
	TO cite this article:WU Rui-Jia,ZHANG Hong-Gang. SOFT-AlignUNet: A Lightweight Transformer with Feature Alignment[OL].[ 4 January 2022] http://en.paper.edu.cn/en_releasepaper/content/4755984


	6. Data Race Detection Method Based on Parallel Data Flow Analysis
	GE You,JIN Dahai
	Computer Science and Technology 30 December 2021
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (483K B)

	Abstract:Data race faults are the most common and difficult to debug faults in parallel programs, and they have characteristics such as multiple inconsistencies in running results, and the results of dynamic test detection are often very poor. According to the static analysis method, the data race in the project is detected, and the credibility of the project is improved, which has become a research hotspot in the field of software testing.This paper proposes a data race detection method based on parallel data stream on the parallel program implemented by OpenMP.Parallel program modeling is carried out first, and parallel data flow analysis is performed on the program according to the characteristics of the OpenMP guidance instruction, and the reading and writing relationship of variables in the parallel region is determined to provide data support for subsequent detection; afterwards, the data race model is constructed for the program, and the data race detection algorithm is carried out on the control flow node according to the read-write relationship of the variable, and the data race is detected according to the state change. In the comparative experiment with other data race tools, it is proved that this method is reliable and stable in detecting data race.
	TO cite this article:GE You,JIN Dahai. Data Race Detection Method Based on Parallel Data Flow Analysis[OL].[30 December 2021] http://en.paper.edu.cn/en_releasepaper/content/4756049


	7. Monocular Mobile Cube SLAM Based on Fusion of IMU Initialization
	CUI Xiaoyang,WANG Jing
	Computer Science and Technology 28 December 2021
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (513K B)

	Abstract:With the continuous development of human-computer interaction, the challenging lip reading task has become a research hotspot. As a part of the whole lip reading model, feature extraction module determines the upper limit of the overall ability of the lip reading model from the source. The current lip reading feature extraction model is composed of 2D and 3D convolution layers, which can not meet the strong correlation of local features. In order to solve this problem, this paper first analyzes the shortcomings of existing models based on anatomical principles and basic models; then proposes global features to enhance local features through the interaction of global features and local features; finally, based on the above design ideas, designs a lip reading feature extraction model with local feature enhancement. Through the experimental verification on the data set, the lip reading feature extraction model constructed in this paper has stronger representation ability.
	TO cite this article:CUI Xiaoyang,WANG Jing. Monocular Mobile Cube SLAM Based on Fusion of IMU Initialization[OL].[28 December 2021] http://en.paper.edu.cn/en_releasepaper/content/4755986


	8. When Tiramisu meet WGAN: A Model for Portrait Matting
	ZHANG Jia-xuan,Peng Hai-peng
	Computer Science and Technology 24 December 2021
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (1223K B)

	Abstract:Image matting technology is an important topic in computer vision and has been used in a wide variety of applications. In particular, portrait matting is widely used in film, advertising, short video and other areas. Matting is not a simple image segmentation problem, a good matting algorithm can deal with hair and other details accurately. In this paper, we combine the fully convolutional dense network named Tiramisu with the generative adversarial network(GAN). Through Tiramisu, we can obtain a coarse alpha channel and the RGB channels with chromatic aberration. Next, we use the structure of GAN to refine the details, the network needs to adjust the boundary of alpha channel and the color values of the other three channels. However, the instability is GAN's shortcoming. In order to solve this weakness, Wassertein distance is introduced into the loss function of GAN. It completely solves the problem of unstable training of GAN, and basically solves mode collapsing problem. We compare the proposed approach with several state-of-the-art image matting methods on synthetic datasets. The main difference between our method and other methods is that we generate an RGBA image rather than only an alpha channel. It makes full use of the advantages of GAN in the field of image generation. The result shows our model outperforms most of its competitors in the image quality, and it is superior than all of competitors in real-time performance.
	TO cite this article:ZHANG Jia-xuan,Peng Hai-peng. When Tiramisu meet WGAN: A Model for Portrait Matting[OL].[24 December 2021] http://en.paper.edu.cn/en_releasepaper/content/4755954


	9. Implementation of DSL compiler based on ANTLR
	Jiang Ya-Yun, Man Yi
	Computer Science and Technology 06 December 2021
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (416K B)

	Abstract:This paper presents a method to implement domain specific language compiler based on ANTLR tool. The domain specific language studied in this paper is a chip development language but higher than instruction language. The language is concise and all variables in it refer to registers, The purpose of this paper is to realize the compiler of it and obtain its corresponding assembly code. The implementation of the compiler is based on ANTLR tool, which eliminates the cumbersome process of handwritten lexical analysis and syntax analysis. The listener generated by it is combined with a handwritten Conversion tool to compile the domain specific language. Finally, the results are given to verify the correctness of the compiler.
	TO cite this article:Jiang Ya-Yun, Man Yi. Implementation of DSL compiler based on ANTLR[OL].[ 6 December 2021] http://en.paper.edu.cn/en_releasepaper/content/4755893


	10. Load-Aware Transmission Mechanism for NVMeoF Storage Networks
	Qiao Xinghan,Xie Xuchao,Xiao Liquan
	Computer Science and Technology 21 October 2021
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (641K B)

	Abstract:NVMe over TCP is a key technology for building large-scale high-performance storage systems.It can realize NVMeoF (NVMe over Fabrics) storage network based on the existing data center network infrastructure and standard TCP/IP software protocol stack. This article designs and implements the Load-Aware NVMeoF message processing mechanism LANoT (Load-Aware NVMe over TCP). Firstly, the interrupt merging technology based on aggregated PDU is used to alleviate the interrupt storm problem and achieve high throughput. Secondly, matching the special message processing mechanism, which can effectively improve its key performance indicators for applications according to the I/O characteristics of different dedicated queues . This paper implements the LANoT prototype system in the Linux kernel. The performance test results show that compared to the NVMe over TCP implementation in the standard Linux kernel, LANoT can reduce CPU resource consumption by more than 50% and increase IOPS by more than twice.
	TO cite this article:Qiao Xinghan,Xie Xuchao,Xiao Liquan. Load-Aware Transmission Mechanism for NVMeoF Storage Networks[OL].[21 October 2021] http://en.paper.edu.cn/en_releasepaper/content/4755658

	Check out RSS, or use RSS reader to subscribe this item