Beiqian Dai

University of Science and Technology of China

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Trends

Author Order

Document Type

Co-Authors

Yanlu Xie

Beijing Language and Culture University

Zhiqiang Yao

Anhui University

Xi Zhou

Guangdong Academy of Sciences

Jun Sun

West China Medical Center of Sichuan University

M. Liu

University of Science and Technology of China

Changfu Wang

Henan University

Wei Lu

Beijing Microelectronics Technology Institute

Hui Li

University of Science and Technology of China

Xi Zhou

Institute of Optics and Electronics, Chinese Academy of Sciences

Frank K. Soong

Microsoft Research Asia (China)

Cooperative Institutions

University of Science and Technology of China

Chinese Academy of Sciences

University of Chinese Academy of Sciences

Shanghai Jiao Tong University

Harbin Institute of Technology

Northwestern Polytechnical University

Nanjing University of Science and Technology

Huazhong University of Science and Technology

Microsoft Research Asia (China)

Microsoft Research (United Kingdom)

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Field

Modeling Glottal Source for High Quality Voice Conversion

Jun Sun Beiqian Dai Jian Zhang Yanlu Xie

A novel modeling method for glottal source is proposed for improving the naturalness and quality of synthetic speech. This paper utilizes the high correlation between vocal tract parameters and glottal source to model glottal source. Vocal tract parameters (LSF) are clustered into some classes. Within each class, a LSF vector closest to centroid and its corresponding glottal wave derivative are selected as a code vector representing different phonetic class of voiced speech. At the stage of voice conversion or synthesis, we can find the relevant glottal source by virtual of finding the closest matched vocal tract parameters. Experiment results show that this vocal tract related glottal source model significantly outperform Rosenberg model and LF model. Correlation coefficients between vocal tract related glottal source and original glottal source increase 27% and 30.13%, spectral distance between synthetic speech and original speech reduce 50.5% and 51.48% respectively, comparing with Rosenberg model and LF model

Vocal tract

Centroid

Source model

10.1109/wcica.2006.1713833

Cite

Citations (2)

The influence of fundamental frequency to performance of speaker recognition using MFCC

Wei Lu Beiqian Dai Hui Li Qingsong Liu

Fundamental frequency can affect the accurate description of MFCC and degrade the performance of speaker recognition system,so a new parameter called SMFCC(smoothing MFCC) was proposed.The parameter mainly employs the method of smoothing amplitude spectral to reduce the effect of fundamental frequency.Experiments in YOHO speech database show that SMFCC is more effective and robust than the conventional MFCC,especially in the female speaker recognition system.

Mel-frequency cepstrum

Smoothing

Fundamental frequency

Source

Cite

Citations (1)

Improved covariance modeling for GMM in speaker identification

Interspeech 2022 (2005)

Xi Zhou Zhiqiang Yao Beiqian Dai

Speaker identification

Identification

10.21437/interspeech.2005-669

Cite

Citations (3)

A knowledge-based understanding system for the Chinese spoken language

Yanghai Lu Beiqian Dai

Spoken Language

10.21437/icslp.1990-322

Cite

Citations (0)

Text-independent speaker verification based on GMM all statistical parameters and SVM

Beiqian Dai

An approach for text-independent speaker verification with telephone speech was proposed,based on GMM all statistical parameters and SVM to solve the problem of building SVM speaker models with speech features.By using GMM derived from the GM via adaptation of the means to extract statistical parameters,new parameters were defined based on all statistical parameters of SVM and were used with linear kernel to build SVM speaker models which had good performance.The experimental results of the approaches based on GMM-UBM and another GMM statistical parameters and SVM in the database of NIST05SRE show excellent properties of the proposed.

Speaker Verification

Kernel (algebra)

Statistical Analysis

Source

Cite

Citations (0)

Hierarchical Gaussian mixture model for speaker verification

4th International Conference on Spoken Language Processing (ICSLP 1996) (2002)

Ming Liu Eric Chang Beiqian Dai

Speaker Verification

10.21437/icslp.2002-411

Cite

Citations (19)

Improved Two-stage Wiener Filter for Robust Speaker Identification

2022 26th International Conference on Pattern Recognition (ICPR) (2006)

Yanlu Xie M. Liu Zhiqiang Yao Beiqian Dai

In order to solve the problem of robustness in textindependent speaker identification, two-stage Wiener filter presented in the ETSI standard is used in this paper. However, the performance of the voice activity detection (VAD) block in the standard is not so satisfying under lower SNR. Thus constrainedclustering vector quantization is presented to improve the block in the Wiener filter method. Experiments based on the MSRA's mandarin speech corpora show the good robustness of the improved method to several common additive noises.

Speaker identification

Wiener filter

Identification

Filtering theory

10.1109/icpr.2006.696

Cite

Citations (5)

A fast implementation of factor analysis for speaker verification

Interspeech 2022 (2010)

Qingsong Liu Wei Huang Dongxing Xu Cai Hong-bin Beiqian Dai

Speaker Verification

Factor (programming language)

10.21437/interspeech.2010-142

Cite

Citations (2)

Extended Hierarchical Gaussianization for scene classification

Minqiang Xu Xi Zhou Zhen Li Beiqian Dai Thomas S. Huang

In this paper, we propose a novel image representation for scene classification. Firstly, we model multiple order statistics of image patches via Gaussian Mixture Model(GMM) in a Bayesian framework. Secondly, we combine the information of mean and covariance of the GMM and represent it as a mean-covariance supervector through a new distance metric. Experimental results demonstrate that our new representation, by just using nearest centroid classifier, has significantly outperformed all existing methods on the fifteen scene category database.

Centroid

Contextual image classification

Representation

10.1109/icip.2010.5653825

Cite

Citations (7)

Improved GMM-UBM/SVM For Speaker Verification

Minghui Liu Beiqian Dai Yanlu Xie Zhiqiang Yao

This paper combines Gaussian Mixture Model-Universal Background Model (GMM-UBM) and Support Vector Machine (SVM) through post processing the GMM-UBM scores of different dimension feature parameter with SVM in speaker verification. Because different dimension feature makes different contribution to recognition performance and SVM has good discriminability, this combining approach yields significant performance improvements on decision-making. Experiments on text-independent speaker verification in NIST05 8conv4w-1 conv4w data showed that the actual detection cost function (DCF) of the test system was reduced to 0.0290 from 0.0343.

Speaker Verification

Feature (linguistics)

Feature vector

10.1109/icassp.2006.1660173

Cite

Citations (16)