On my sophomore year in college, a friend convinced me, a then business major student, to take a course in computer science. Translating problems to interactive graphics applications, encountering unfamiliar and challenging scenarios, and finding solutions, the course brought me profound satisfaction. I soon realized that — for I was fascinated by problem-solving and inclined to interpret the world through the lenses of logic, algorithms, and proofs — this was the field I wanted. Therefore, I switched to Mathematics and Computer Science degrees with the determination to spend every summer in college catching up on coursework.

In the beginning, only driven by my personality and enthusiasm, I had been confused and wondering how to combine and make efficient use of both areas in my future career. Later in my working as the director of the costume department for the Indiana University Heping Chinese Theater Club, I found that lots of non-Chinese-speaking audience members had to read our printed translations of the lines. That caused me to contemplate—can I develop a more widely-used simultaneous translation utility that is specifically for art purpose, such as drama?

As both a fan of drama and music and a student majored in math and CS, I hope I can apply my classroom knowledge to make some practical progress in facilitating arts, especially providing easier access for Chinese drama. Audio technologies powered by machine learning are widely used to improve practical devices such as cell phones and hearing aids. We can also apply them to enhance music, theater, and other arts. Some critics argue that songs and paintings generated by AI mobile applications replace hu[University Name] creativity and cannot be considered art. But the patterns and predictions made by AIs expand the capabilities of hu[University Name] artists and enhance the enjoyment of audiences, instead of simply replacing them. Another source of anxiety surrounding AI is that it will leave workers out of jobs. In reality, it can enrich lives by relieving people from menial or duplicated forms of labor and freeing up time for more creative and fulfilling pursuits. Believing in AI’s promising future, I make my long-term career goal as a machine learning researcher and becoming a leader in the team.

To start my path, I enrolled in a machine learning course with Professor Donald S. Williamson, whose primary research area was speech processing. I applied to assist him with a research project to practice concepts from lecture. It was based on the paper “Boosting Contextual Information for Deep Neural Network Based Voice Activity Detection” by Dr. Deliang Wang and Dr. Xiaolei Zheng. During the research, the biggest challenge for me was the code used was python with TensorFlow, which I was new to. To learn the new language and framework, I looked through the guides from the TensorFlow website and self-learned it by reading articles and watching tutorials for different types of neural network, such as DNN, CNN, and LSTM. If there were errors that I cannot fix on my own, I looked for help from other graduate students supervised by the same professor during or after our weekly report meetings. To maintain an adequate running time, I asked Professor Williamson for sponsoring access of IU Big RedII server and ran all my test with reduced the training batch size from 128 to 64. I also increased my result AUCs from around 50% to desired 84.5% by changing training and testing with the same type of sound but different types of noise to training and testing both with different types of sound and noise. In addition to a more practical understanding of deep learning concepts, I improved my adaptability and problem-solving skills from immersing in a research group environment as well.

Northwestern University’s M.S. in CS program is ideal to help me achieve my career goals. First, the Interactive Audio Lab, specialized in audio processing, strongly interests me due to its goal to produce a practical solution to make art accessible to all through machine learnings. I read several papers of Dr. Bryan Pardo who is a primary researcher in the lab. The “Music/Voice Separation Using the 2D Fourier Transform” especially inspired me. I learned fundamental Fourier Transform in my complex analysis course but never thought it would have large impact on speech separation. Also, it exposed me to the concept of filtering periodic voice out of the non-periodic without knowing the period. In addition, Dr. Pardo teaches [University Name]y interesting courses that are uncommon in my analysis comparing different schools’ curriculum, such as Machine Learning Perception for Music & Audio, Digital Musical Instrument Design, and Computational Auditory Scene Analysis. For example, Machine Learning Perception for Music & Audio weights projects in a large percentage without having any test. It examines students by making practical research of theories instead of memorizing. I prefer more projects from these course as they will expose me on things deeper than what was exactly taught in the classroom. I wish to finish my program in the project or thesis plan with a topic in audio processing supervised by Dr. Bryan Pardo and assist in the Interactive Audio Lab as my next step.

Seeing people got surprised when they knew my majors was nothing new. But there was once a uber driver, whose wife worked in the CS field, said: “She’s been so anxious about female participants are way too less than it could be, for [University Name]y reasons. But I will tell her your story, I’ll let her know there’re [University Name]y like her working to make a difference, and I’m sure she will be happy. I’m sure, girl.” I felt a quaver in his voice. “She will.” I looked firmly at him. I knew the percentage of women in CS study or related working field was very low, not to mention the leader researchers. As a female, I strive for a leading position in researchers and wish my experience could encourage more to confront the bias that female can’t do CS or math. I remain grateful of my friend’s suggestion that accidentally helped me discover my true passion and path.