The three of us: Jiahong Zhao, Christian Ritz and Jiangtao Xi

UOW Phd candidate, Jiahong Zhao, with the help of supervisors Christian Ritz and Jiangtao Xi, is creating new approaches to signal processing algorithms and microphone placement, which will hopefully help create the next generation of home assistant products, automated vehicles, robots and other smart devices that rely on human-machine interaction.

Meet the Candidate: Jiahong Zhao

Born and bred in northeast China, in the city of Jilin, Jiahong moved to Beijing to complete his Bachelor and Masters degrees in Engineering at the University of Science and Technology Beijing (USTB), majoring in electrical engineering. After working as an electrical engineer for five years in companies such as General Electric, China Aerospace Science and Technology Corporation (the Chinese NASA) and Meituan (the Chinese Facebook), Jiahong decided to pursue the highest qualification of Doctor of Philosophy.

Pictured: Jiahong Zhao in his acoustics laboratory.

Thesis title

Co-Prime Microphone Arrays: Geometries, Beamforming and Speech Direction of Arrival Estimation

Can you describe your research topic?

The recording of sound in the real world is challenging. Because interfering sound such as noise and echo is everywhere, humans’ speech can be hard to hear by others or machines and needs to be purified and enhanced. If the received speech is not improved, smart products including smart home assistants and autonomous vehicles would only work in the laboratory.

My research designs and tests new methods of sound recording using multiple microphones. I have designed a new specific layout that uses a small number of microphones to obtain equivalent performance to that of a much larger number of microphones using traditional layouts. For example, the new layout with 16 microphones can be equivalent to a traditional layout with 72 microphones. By using my signal processing algorithms, the new layout records sound only from one desired direction, whilst suppressing sound from all other directions. As a result, my research significantly reduces the cost and size of using multiple microphones, whilst improving the speech recording quality in practical smart products.

Where does your interest in this field come from?

My interest in signal processing was fostered in my undergraduate studies and my time spent working in industry, where I developed a few projects on audio information hiding, image processing, robotics and autonomous vehicles.

When I started my PhD study, my principal supervisor, Prof. Christian Ritz, proposed a few possible research directions, which were highly related to my background. During the first six months, I read as much as I could and determined the research gap I would investigate, one that was both interesting to me and of significance to the field. After reviewing the relevant literature and trying out a few initial ideas, Christian and I selected my research topic together, which made me feel inspired and supported.

How did you find your supervisor?

Finding Christian as my supervisor was a little destined. When planning to study in Australia, I did a thorough read of many researchers’ profiles on the Internet but could not really understand all the details. I sent enquiry emails to a few researchers whose work I found could be relevant to my background. Unexpectedly, only two of them replied. One researcher kindly told me that my email was in the junk email due to the uncommon mail server, and he was nearly retired. The other was Christian, which might mean that few researchers including Christian received my email. As his research exactly matched my interest, and his nice personality shone through in our communications, I decided to do research with him without any hesitation. Later on, after I enrolled in my PhD study, Christian introduced Jiangtao as my co-supervisor, who I found was one of the top experts in my research field. I was more than happy to have him in the supervisory team.  

What’s been the most exciting part of your journey so far? What’s been the most challenging?

While having publications is exciting to all PhD students, I would say the most exciting experience of my journey was participating in competitive award applications and contests, especially for those happened during the pandemic that hindered participation.

From the global visibility of my PhD research in the first IEEE 5-Minute Video Clip Contest to the last-minute submission to the UOW Global Challenges Video Contest.

From the 24 hours wearing masks in the IEEEXtreme programming hackathon to the votes from my parents and international experts in my field in the UOW 3-Minute Thesis Competition, winning the People’s Choice Award.

Last but foremost, my quick pivot to remote teaching was recognised in 2021 with a Vice-Chancellor’s Award for Outstanding Contributions to Teaching and Learning (OCTAL). These efforts and resilience to hurdle the difficulties turn out enjoyable and rewarding, which polished my research skills and mindsets indirectly. I believe the dots will be connected in my future career, which are beyond winning the awards themselves.

The most challenging situation is to conduct research under COVID. It is not only about the lack of professional equipment, lab resources, supports, etc., but mainly about coping with uncomfortable psychological feelings that might occur along the journey. The uncertainty of the potential challenge during the pandemic is the most challenging matter that one has to tackle.

Looking back on the past few years, the feeling is mostly like riding a roller coaster, but I believe it has been rewarding: the increase in knowledge, the polished methodology of doing research and the stronger mindset and skills in tackling challenges. These benefits can be applied to any pursuits in life and can lead to a fulfilling, bright future. Furthermore, the achievements along the journey can contribute to the wider community and have a positive impact on others, with any luck. For all of these reasons, pursuing a PhD degree is worthwhile.

How do you think your research will make a difference?

My research directly brings clearer recordings of speech by modifying the layout of microphones in existing smart products, such as Amazon Echo and Google Home. By applying my designed signal processing algorithms to the new layout, the smart products can clearly communicate with humans and other machines in real-world noisy environments and thus become more practical. This makes the human-machine interaction possible and enables the usage of AI techniques in the machines. Hence, my research can lead us to a smart future by making AI hear us much better.

What’s next?

I hope to continue my PhD research in audio and speech signal processing using multiple microphones after graduation. I also intend to utilise my previous strength in embedded system to develop research on the currently heatedly discussed topic artificial intelligence of things (AIoT) and apply my outcomes in audio and speech signal processing to provide additional cues to AIoT solutions.

I am currently an associate research fellow in both SECTE and SMART at UOW, which provide valuable platforms to allow me to share my passion and skills in those research directions.

What advice would you give someone considering doing postgraduate studies?

Make sure that the program to be studied closely aligned to your passion or interest. This passion will give you an infinite energy source to help to go through the up-and-downs along the journey. Find a supervisor (or two) who well matches your background and personality, because he/she is the person you will most frequently communicate with, and obtain knowledge, support and inspiration from during the next few years.

Keep your own pace: as a marathon runner, I do find that postgraduate study shares similar philosophies to running a marathon. Focus on your own work, move forward step-by-step and persist, you will eventually arrive at your destination.  

Meet The Supervisors

Professor Christian Ritz

Professor Christian Ritz’s research interests are in the area of signal processing for speech, audio, acoustics and visual information. His current projects include: microphone array signal processing for enhancing sound recorded in classrooms; sound scene classification for dementia care environments; spatial audio for Virtual and Augmented Reality; and undersea surveillance. These projects are funded by the Australian Research Council, overseas research institutes and defence organisations. Past projects include audio data compression for next generation 3D teleconferencing, miniature microphone arrays for hands-free communication and automatically annotating “speaker turns” in meeting recordings and Quality of Experience (QoE) for mobile multimedia applications.

Professor Ritz is a member of the Centre for Signal and Information Processing (CSIP), part of the Signals, Information and Communications Research Institute (SICOM), a research powerhouse in the Faculty of Engineering and Information Sciences at the University of Wollongong.

Professor Jiangtao Xi

Professor Jiangtao Xi has been the Head of School in the School of Electrical, Computer and Telecommunications Engineering at UOW for more than 10 years and was appointed as the Associate Dean Global Engagement in the Faculty of Engineering and Information Sciences at UOW in early 2022. After completing his PhD at UOW in 1996, Jiangtao took a Postdoctoral Fellow position at the Communications Research Laboratory, McMaster University, Ontario, Canada and became a Member of Technical Staff at Bell Laboratories, Lucent Technologies Inc., NJ, USA. He held the position of Chief Technical Officer at TCL IT Group Co. in China from 2000 to 2002, after which he returned to UOW in 2003.

Professor Xi’s research interests include signal processing and its applications in various areas such as instrumentation and measurement, as well as telecommunications.

How does Jiahong’s research relate to your own research?

Jiangtao: My research interests are mainly in the areas of signal processing and applications to telecommunications, multimedia (e.g., image, audio and video), sensing and measurement.  These include Jiahong’s research on audio signal processing.

Christian: Jiahong’s research directly relates to my research interests in signal and information processing for speech, audio and acoustic applications. In particular, his project researching new ways of recording sound using multiple microphones so that the signal is of higher quality compared to using a single microphone. One of the applications of this research is for hands-free communication devices, such as products like Google Home, where a user can simply speak to their device to play some music, stream a new TV show or even start an online Zoom meeting. This relates to a number of other projects I have conducted over many years to find out better ways of recording 3D sound so that it is clearer and can provide more accurate information about sound sources and their location.

Can you tell me about Jiahong’s success so far?

Jiangtao: Jiahong has contributed a lot to the area of his research.  He proposed a set of new structures for microphone arrays and developed novel algorithms for achieving optimal performance, and the results have been presented or published on top international conferences and journals.

Christian: Jiahong has developed a number of new approaches to designing microphone arrays that require few microphones to achieve the same quality of recording that normally requires many more microphones. This provides an advantage in terms of size and cost for installing these in hands free communication devices. The focus has been on the signal processing algorithms that have been designed to jointly process multiple microphone recordings to form a highly focused recording of a target sound, such as a person speaking, while removing unwanted sounds from other sources, such as other people talking or background noise in a room. These signal processing algorithms are basically mathematical algorithms that are implemented as computer programs to process the digital signals captured by the microphones. His research has been published in a number of international conferences and he has also received some awards for his research, including from a video contest conducted by a major international conference as well as the annual three minute thesis competition.  

What makes a successful PhD candidate?

Jiangtao: A successful PhD candidate is self-motivated, able to focus and persistently work on the research problem.

Christian: A successful PhD candidate needs to be curious about finding new ideas and approaches to solving problems; passionate about their research project; dedicated and committed to achieving outcomes and a strong ability to plan their research. The PhD journey is not always easy and it requires a certain amount of resilience and persistence to push through the more challenging times—research is not always predictable and doesn’t always result in the outcome you are looking for. While our discipline is quite technical, a key skill that PhD students develop is how to communicate clearly and accurately, particularly when writing papers but also when presenting work at conferences or other occasions. This is something that PhD candidates and their supervisors spend a lot of time on but the rewards are papers accepted for publication and ultimately a great result from their thesis examination.

How do you guide candidates on their journey?

Christian: In the early stages it is about encouraging the candidates to read as much as they can of the relevant literature and also to try out ideas early on. In the research we conduct, this often means trying out some existing signal processing algorithms in software using some existing implementations that might be available. “Learning by doing” is a good way of becoming skilled in the area of signal processing. It’s important at these early stages to also develop a good set of potential research questions and goals that can be narrowed down to form part of their first year research proposal. As students move into their second year I encourage them to continue trying out some new ideas and building on their early successes as well as aiming to write papers that can be submitted to conferences, which is a great way of helping to understand whether or not the ideas they have are novel enough and achieve results that are worth of publication. As the candidate moves into year 3 they should be able to more confidently lead their research, showing more of their own initiative with ideas, and be able to write papers to a high standard without requiring as many edits as with their earlier publications. So a key aspect is continually offering encouragement and support.

Jiangtao: Rather than giving “instructions”, I believe it is more important and effective to share with them my experience (failure and success), knowledge and skills. A good supervisor treats PhD students as friends and teammates, helping, motivating and guiding students to be successful in their research.

How should potential PhD candidates get started in securing a PhD position?

Christian: Getting in contact with potential supervisors who research in their area of interest is one of the first steps. At first, students might not know what areas might interest them and so it is a good idea to try and talk to as many potential supervisors as possible to learn more about their research areas. Spending some time to read some of their papers is also worthwhile. If possible, speaking to some existing PhD students with the same supervisor can be helpful as well. Potential students should remember that a PhD is at least 3 years, so working out if they feel they can work well with the supervisor and potentially also other researchers in their research group is a good idea as it will help make the time much easier.

What’s the best thing about being a supervisor?

Christian: Being a supervisor is an important part of being a researcher. As a supervisor you get to work really closely with your students on problems that you are both very interested in. Both you and your students get to learn about new things and develop new knowledge and often are able to co-author papers together. Seeing a PhD student go from a very early stage researcher to someone who is a successful PhD student and graduate is very rewarding.

In most cases I have supervised international students and it is often the case that their PhD is the first time they have travelled to Australia – it is rewarding to see how they develop a life in Australia. It’s also rewarding to see my PhD graduates go on to successful careers after graduating, which for me has included research roles in Universities or industry roles at top companies like Dolby and Facebook.

If you are interested in speaking to Jiahong about his research or his experiences as a UOW PhD student, get in contact:

Email: jiahong_zhao@uow.edu.au 

Twitter: @Zhao_Jiahong

LinkedIn: View Jiahong Zhao's profile