Computer Science Signal Processing

Xiaojia Zhao

*** ****** **********, **** **** Avenue

Columbus, OH, 43210

Cell Phone: 614-***-****

Email: ***********@*****.***

Homepage: http://www.cse.ohio-state.edu/~zhaox

Objective and Research Interests

Ph.D. candidate at Department of Computer Science and Engineering at the Ohio State University.

Looking for summer internship of 2013. Research interests include Artificial Intelligence, Robust

Speaker/Speech Recognition, Speech Separation, Speech/Signal Processing, Machine Learning and

Computational Auditory Scene Analysis (CASA).

Summary of Skills

Programming Languages: Matlab, C/C++, Java, C#

Operating Systems: Linux and Windows

Academic Background: Artificial Intelligence, Machine Learning, Speaker/Speech recognition,

Digital Signal Processing, Computer Architecture, Operating Systems, Algorithms Analysis and

Design, Probability Theory, Linear Regression, Applied Multivariate Analysis

Work Experience

Graduate Teaching Associate

March, 2012 - current

The Ohio State University Columbus, OH

Help students major in business school develop problem-solving skills using Microsoft Excel, Access, etc.

Responsible for coordinating all the TAs of the same class

Graduate Research Associate

September, 2008 to March, 2012

The Ohio State University Columbus, OH

Work with Professor DeLiang Wang in the research area of single microphone source separation and its

application on speaker recognition using machine learning techniques.

Projects

Noise-robust Speaker Recognition

Designed a robust speaker identification system that works reasonably well in a variety of noisy conditions.

We first propose a novel speaker feature, gammatone frequency cepstral coefficients (GFCC), based on an

auditory periphery model and show that this feature captures speaker characteristics and performs

substantially better than conventional speaker features under noisy conditions. To deal with noisy speech,

we apply CASA separation and then either reconstruct or marginalize corrupted components indicated by a

CASA mask. The complementary advantages of the two methods are utilized by a combination to achieve

further performance improvement.

Page 1 of 2

Analyzing Noise Robustness of GFCC and MFCC Features in Speaker Identification

In the previous project, the proposed GFCC feature exhibits superior noise robustness to the well-known

mel-frequency cepstral coefficients (MFCC). To gain a deep understanding of this, we conduct this project.

We first analyze all of their differences, which helps us to generate a number of hypotheses. For each

hypothesis, we design a corresponding set of experiments to test it. In this way, we are able to narrow

down possible explanations, which eventually reveal the desired answer.

Speaker Recognition in Noisy and Reverberant Conditions

In real world acoustic environments, speech often occurs simultaneously with noise and reverberation. Our

previous projects mainly focus on noise robustness of speaker recognition systems. We conduct this project

to explore the joint effects of noise and reverberation. We first train speaker models in selected noise-free

reverberant conditions. We then tackle noise from two different perspectives, bounded marginalization and

direct masking, based on source separation results from a deep neural network. The proposed system that

combines the two substantially improves speaker identification performance compared to related systems

in a wide range of reverberation time and signal-to-noise ratios.

Education

The Ohio State University Columbus, OH, USA

09/2008 06/2014

Ph.D. in Computer Science and Engineering (GPA: 3.94 / 4)

09/2008 06/2012

M.S. in Computer Science and Engineering (GPA: 3.94 / 4)

Major: Artificial Intelligence

Minor: Signal Processing and Statistics

Nankai University Tianjin, China

Bachelor in Engineering (GPA: 3.85 / 4. Rank 1st / 152) 09/2004 06/2008

Major: Software Engineering (Graduated with Distinction)

Honors and Awards

First-Class Student Scholarship of Nankai University

Excellent Student Leader Award of Nankai University

Outstanding High School Student Award of Henan Province of China

Publications

X. Zhao, Y. Shao, and D.L. Wang: Robust speaker identification using a CASA

front-end. Proceedings of ICASSP-11, pp. 5468-5471, 2011.

A. Narayanan, X. Zhao, D.L. Wang, and E. Fosler-Lussier: Robust speech recognition using

multiple prior models for speech reconstruction. Proceedings of ICASSP-11, pp. 4800-4803, 2011.

(The first two authors have equal contribution).

X. Zhao, Y. Shao and D.L.Wang, "CASA-Based Robust Speaker Identification," IEEE Trans.

Audio, Speech and Language Processing, vol.20, no.5, pp.1608-1616, 2012.

Resume of Xiaojia Zhao Page 2 of 2

Contact this candidate