检索结果-内蒙古大学图书馆

3rd International Conference on Pattern Recognition and Machine Learning, PRML 2022

作者： Zeng, Ailing Ablimit, Mijit Hamdulla, Askar School of Information Science and Engineering Xinjiang University of China China Xinjiang Key Laboratory of Signal Detection and Processing Urumqi830017 China Xinjiang Key Laboratory of Multilingual Information Technology China

ISBN: (数字)9781665499507

ISBN: (纸本)9781665499507

Language identification process is necessary for the downstream tasks of cross lingual and multi-lingual application. This paper proposes a language recognition method based on ResNet-CA-BiGRU-Attention model. Firstly, the speech signal is converted into the corresponding spectrogram, and then the residual network is used for feature extraction according to the distribution of speech energy on the spectrogram. Coordinate Attention and Attention are used to highlight corelated features and suppress irrelevant features. In this paper, five languages are selected from the Common Voice data set. Experimental results show that the proposed language recognition method based on ResNet-CA-BiGRU-Attention model can effectively improve the performance of language recognition. © 2022 IEEE.

关键词： Natural language processing systems

来源：评论

学校读者我要写书评

暂无评论

Speech Separation and Emotion Recognition for Multi-speaker Scenarios 3

Speech Separation and Emotion Recognition for Multi-speaker ...

引用

3rd International Conference on Pattern Recognition and Machine Learning, PRML 2022

作者： Jin, Rong Ablimit, Mijit Hamdulla, Askar School of Information Science and Engineering Xinjiang University of China China Xinjiang Key Laboratory of Signal Detection and Processing Urumqi830017 China Xinjiang Key Laboratory of Multilingual Information Technology China

ISBN: (数字)9781665499507

ISBN: (纸本)9781665499507

Emotion recognition in multi-speaker scenarios has been a hot topic of continuous research. In the single-channel overlapped speeches, the recognition accuracy is greatly degraded compared to clean speeches. Therefore, a speech separation task for multi-speaker separation at the front end of emotion recognition is necessary. The overall efficiency is the combined effect of separating a single speech from overlapped voices and emotional recognition. The speech separation experiments in this paper are based on Conv-TasNet, using the WSJ0-2mix dataset as training data to obtain the trained separation model;meanwhile, the speech emotion recognition experiments are inherited from Wav2vec-2.0 pre-trained model with a multi-task learning framework to recognize both text and emotion. Through the experiments, the possibility of bridging the two tasks and the usability of the separation and recognition results for realistic scenarios are demonstrated, and subsequent work can be carried out on improving the robustness and application to other realistic scenarios. © 2022 IEEE.

关键词： Emotion Recognition

来源：评论

学校读者我要写书评

暂无评论

Research of Pitch Controllability Based on LPCNet 3

Research of Pitch Controllability Based on LPCNet

引用

3rd International Conference on Pattern Recognition and Machine Learning, PRML 2022

作者： Gao, Wenyu Hamdulla, Askar Yang, Xipeng Chen, Yunlin School of Information Science and Engineering Xinjiang University of China China Xinjiang Key Laboratory of Signal Detection and Processing Xinjiang Urumqi China Xinjiang Key Laboratory of Multilingual Information Processing China Mobvoi Ai Lab Suzhou China

ISBN: (数字)9781665499507

ISBN: (纸本)9781665499507

With the development of speech synthesis technology in most scenarios, modifying the pitch of audio has been widely used. Currently most pitch controllable methods are achieved by using a separate reference encoder neural network, but this approach requires training more complicated neural networks and does not generalize to unseen speakers without speaker adaptation. To address this problem, a method based on LPCNet vocoder for explicitly pitch control is proposed. Firstly, during the training phase, by optimizing the training features, the model can synthesize more natural and robust voices. Next, the fundamental frequency is adjusted by inputting a control curve in the inference stage, and then the acoustic features are inferred into waveform using the vocoder LPCNet. The experimental results show that the proposed method can control the pitch flexibly and the sound quality of the synthetic voice is improved. © 2022 IEEE.

关键词： Speech synthesis

来源：评论

学校读者我要写书评

暂无评论

Research on Text Recognition of Natural Scenes for Complex Situations 3

Research on Text Recognition of Natural Scenes for Complex S...

引用

3rd International Conference on Pattern Recognition and Machine Learning, PRML 2022

作者： Yu, Wenhua Ibrayim, Mayire Hamdulla, Askar Xinjiang University School of Information Science and Engineering Urumqi China Xinjiang Key Laboratory of Signal Detection and Processing Urumqi China Xinjiang Key Laboratory of Multilingual Information Technology Urumqi China

ISBN: (数字)9781665499507

ISBN: (纸本)9781665499507

As a great invention that affects human development and progress, writing plays an essential role in human life and learning and the inheritance and development of culture. For thousands of years, the inheritance of Chinese culture has been inseparable from the writing records. Papermaking and printing have played a crucial role in the development of the times. Text extraction and recognition technology in natural scenes, that is, text recognition from photos or videos, has become a popular research direction in computer vision in recent years. Scene text recognition is a fundamental task in computer vision. It is often a critical step in many text-related tasks, such as document retrieval, card recognition, license plate recognition, and other tasks. The vital difference between the text recognition process and text detection is that the text in the photo or video needs to be recognized. detection and recognition are mainly divided into two steps: locating text areas in photos or videos, that is, finding words or text lines of the bounding box;then the positioned text is recognized. Combining these two steps results in End-to-end Recognition of text. The continuous development of scene text detection and recognition algorithm system will lay the foundation for the research of recognition problems such as multilingual recognition of scene text and formula recognition. © 2022 IEEE.

关键词： Computer vision

来源：评论

学校读者我要写书评

暂无评论

Speech Language Identification Using CNN-BiLSTM with Attention Mechanism 3

Speech Language Identification Using CNN-BiLSTM with Attenti...

引用

3rd International Conference on Pattern Recognition and Machine Learning, PRML 2022

作者： Yilihamu, Dilizhati Ablimit, Mijit Hamdulla, Askar School of Information Science and Engineering Xinjiang University of China China Xinjiang Key Laboratory of Signal Detection and Processing Urumqi830017 China Xinjiang Key Laboratory of Multilingual Information Technology China

ISBN: (数字)9781665499507

ISBN: (纸本)9781665499507

Aiming at the problems of low feature extraction ability and low recognition rate in language identification task, this paper proposes CNN-BiLSTM language identification method based on attention mechanism from the two aspects of feature and model. In the feature extraction stage, we compare the recognition effects of different speech features such as MFCC, LPCC and spectrogram, Experiments show that the recognition effect of spectrogram features is better in the actual environment. This method extracts the spectrogram of speech data, uses convolution network to obtain the visual features of spectrogram, and introduces attention mechanism, which can pay attention to the language related information in speech fragments and ignore non language information. This method is tested on two cross-channel data sets (AP19-olr) and (AP20-olr) of the Oriental Language Recognition competition. Compared with the language identification methods such as the baseline system x-vector, the equal error rate (EER%) on the two data sets is reduced by 3.38% and 4.66%, The results clearly show that the CNN-BiLSTM language identification method based on attention mechanism effectively improves the feature extraction ability in language identification task, and improves the performance of language identification in practical application environment. © 2022 IEEE.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Language Identification in Overlapped Multi-lingual Speeches 3

Language Identification in Overlapped Multi-lingual Speeches

引用

3rd International Conference on Pattern Recognition and Machine Learning, PRML 2022

作者： Aysa, Zuhragvl Ablimit, Mijit Hamdulla, Askar School of Information Science and Engineering Xinjiang University of China China Xinjiang Key Laboratory of Signal Detection and Processing Urumqi830017 China Xinjiang Key Laboratory of Multilingual Information Technology China

ISBN: (数字)9781665499507

ISBN: (纸本)9781665499507

Speech, languages, and background voices are overlapped in the multi-speaker environment. The speech separation process is necessary for downstream applications like language and speaker identification and ASR tasks. A combined speech separation and language recognition approach is proposed in this paper. First, the monolingual information is extracted from a multi-speaker multilingual speech scenario, then fed the separated speech to a language identification model. The speech separation model is trained in a single-channel overlapped speech data, and the language identification model is trained in Oriental language corpus. The combined effect of these two models is evaluated in the task of language identification. When the language recognition network is fed with the audio spectrograms separated by the source separation network, the Chinese and Korean language recognition results are significantly improved in terms of accuracy and recall rate than the mixed audio spectrograms. © 2022 IEEE.

关键词： Source separation

来源：评论

学校读者我要写书评

暂无评论

Exploration on the mixed practice teaching methods of team experience under the training of new engineering talents

Exploration on the mixed practice teaching methods of team e...

引用

2020 International Conference on Information Science and Education, ICISE-IE 2020

作者： Fei, Shi Hong, Jia Zhen Yun, Zhou Yan Juan, Chen Xinjiang University School of Information Science and Engineering Key Laboratory of Signal Detection and Processing No.666 Shengli Road Tianshan District Urumqi Xinjiang China

ISBN: (纸本)9781665422611

Facing the requirements of the first-class discipline of electronic information engineering and the construction of new engineering courses, this paper analyzes the characteristics of practice teaching under the background of new engineering, puts forward a team experience hybrid practical teaching method for new engineering talents training, expounds the implementation process of this method in detail, and applies the method to the practice project design course of electronic information engineering experimental class. The results show that this method can improve students' self-learning ability, knowledge application ability, problem-solving ability, and engineering practice ability. © 2020 IEEE.

关键词： Curricula

来源：评论

学校读者我要写书评

暂无评论

Network Intrusion Traffic detection Based on Feature Extraction

引用

Computers, Materials & Continua 2024年第1期78卷 473-492页

作者： Xuecheng Yu Yan Huang Yu Zhang Mingyang Song Zhenhong Jia College of Information Science and Engineering Xinjiang UniversityUrumqi830046China Network Department China Mobile Communications Group Xinjiang CoLtd.UrumqiUrumqi830011China Xinjiang Signal Detection and Processing Key Laboratory Xinjiang UniversityUrumqi830000China

With the increasing dimensionality of network traffic,extracting effective traffic features and improving the identification accuracy of different intrusion traffic have become critical in intrusion detection systems(IDS).However,both unsupervised and semisupervised anomalous traffic detection methods suffer from the drawback of ignoring potential correlations between features,resulting in an analysis that is not an optimal ***,in order to extract more representative traffic features as well as to improve the accuracy of traffic identification,this paper proposes a feature dimensionality reduction method combining principal component analysis and Hotelling’s T^(2) and a multilayer convolutional bidirectional long short-term memory(MSC_BiLSTM)classifier model for network traffic intrusion *** method reduces the parameters and redundancy of the model by feature extraction and extracts the dependent features between the data by a bidirectional long short-term memory(BiLSTM)network,which fully considers the influence between the before and after *** network traffic is first characteristically downscaled by principal component analysis(PCA),and then the downscaled principal components are used as input to Hotelling’s T^(2) to compare the differences between *** datasets with outliers,Hotelling’s T^(2) can help identify the groups where the outliers are located and quantitatively measure the extent of the ***,a multilayer convolutional neural network and a BiLSTM network are used to extract the spatial and temporal features of network traffic *** empirical consequences exhibit that the suggested approach in this manuscript attains superior outcomes in precision,recall and F1-score juxtaposed with the prevailing *** results show that the intrusion detection accuracy,precision,and F1-score of the proposed MSC_BiLSTM model for the CIC-IDS 2017 dataset are 98.71%,95.97%,and 90.22%.

关键词： Network intrusion traffic detection PCA Hotelling’s T^(2) BiLSTM

来源：评论

学校读者我要写书评

暂无评论

Towards tiny object tracking for low-illumination wide-field

Towards tiny object tracking for low-illumination wide-field

引用

2022 International Conference on Computer Application and Information Security, ICCAIS 2022

作者： Xie, Zhaodong Jia, Zhenhong Jiang, Wangxi College of Information Science and Engineering Xinjiang University Xinjiang Urumqi830046 China Key Laboratory of Signal Detection and Processing Xinjiang University Xinjiang Urumqi830046 China

ISBN: (纸本)9781510663459

Night is an inevitable scene for surveillance video. Due to the high image resolution, complex background, uneven illumination, and similarity between the target and the background of hawk-eye surveillance video, it is difficult for previous trackers to apply the tracking of a tiny object in such scenes. In this regard, this paper proposes to combine an online automatically and adaptively learning spatio-temporal regularized tracking algorithm with an efficient and effective low-light image enhancement algorithm to improve tracker performance. We constructed a new benchmark that includes 41 night surveillance sequences captured by Hawk-Eye cameras at night. Exhausted experiments have been conducted on this dataset, and the results show that by combining the two methods, the original algorithm can obtain better results in this dataset, and can meet the real-time object tracking, which contributes to the application of tiny object tracking in eagle-eye surveillance video at night. © 2023 SPIE.

关键词： Image resolution

来源：评论

学校读者我要写书评

暂无评论

Research on Speech Synthesis Based on Prosodic Controllability 5

Research on Speech Synthesis Based on Prosodic Controllabili...

引用

5th International Congress on Human-Computer Interaction, Optimization and Robotic Applications, HORA 2023

作者： Gao, Wenyu Hamdulla, Askar Yang, Xipeng School of Information Science and Engineering Xinjiang University of China Xinjiang Key Laboratory of Signal Detection and Processing Xinjiang Urumqi China Mobvoi AI Lab Suzhou China

ISBN: (纸本)9798350337525

In today's highly interactive human-computer world, speech synthesis is widely used in many scenarios, and the requirements for rhyme effects in speech synthesis technology are increasing, so rhyme-controllable models have become a research hotspot. Most current rhyme controllable models are achieved by using a separate neural network to generate reference features, but this approach requires the training of more complex neural network models and the availability of reference audio to achieve display rhyme control. This paper proposes a rhyme-controllable solution based on an end-to-end acoustic model to address the problem of models being unable to precisely control the tone at the word level. The proposed model includes a tone-controllable module, which obtains duration information through the MFA alignment tool and adjusts the tone of the words by using word-level pitch control values and duration information. The acoustic model in this paper is improved by introducing pitch control during the generation of acoustic features and generating more robust audio by combining it with the decoder. In addition, to adjust the overall tone of the audio, a fixed coefficient is multiplied to the pitch values of all frames. Furthermore, this paper also proposes a 48KHz ultra-high-definition audio model by increasing the spectral parameter dimensions and upsampling by a factor. © 2023 IEEE.

关键词： Speech synthesis

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：