School of Engineering Department of Electronic and Computer Engineering 139 Projects in Audio Signal Processing Supervisor: CHAU Kevin / ECE Student: LUI Sin Hang / ELEC SO Chun Hin / COMP Course: UROP 1000, Summer UROP 1000, Summer Source separation is the act of splitting a mixed audio signal into its original sources. Mixing audio signals from multiple sources is an easy process, but the reverse problem, source separation, is a hard problem. Several core ideas on audio signal processing are first given as a brief introduction, such as discrete Fourier transform and its inverse, its properties, and short-time Fourier transform. Their relations to source separation are then discussed. Our ideas on performing source separation based on discrete Fourier transform are also discussed. Lastly, current and potential applications of source separation and how it relates to our upcoming audio processing project will be explored. Projects in Audio Signal Processing Supervisor: CHAU Kevin / ECE Student: ZHANG Luoyi / COSC Course: UROP 1000, Summer This report examines the theoretical and the practical aspects of audio signal processing, with a further focus on its application in the field of deep learning. It covers concepts such as spectral analysis, noise suppression, feature extraction, and their implementations in automatic speech recognition. The study highlights the crucial role of a strong mathematical foundation and broad interdisciplinary knowledge in mastering these techniques. By integrating theoretical learning with practical applications, the report describes how these concepts are used for developing audio processing systems. Through self-reflection, the report underscores the value of perseverance and iterative optimization in solving complex problems. Nanostructured Chemical Sensors and Sensor Systems Supervisor: FAN Zhiyong / ECE Student: LEONG Hok Weng / DSCT Course: UROP 1100, Summer This research aimed to develop a reliable system for locating dangerous gas sources in the environment, crucial for safety and health. By using sensor arrays, regression models, and mathematical analysis, the study explored an approach to gas source detection. Results indicated that at close distances and in the windstable area, the system could predict the angle within a 20 degree margin of error. While this study offered insights into gas source localization approaches, the accuracy of the system indicates the need for further advancements to develop more reliable and precise technologies for detecting and positioning hazardous gas sources in real-world environments.
RkJQdWJsaXNoZXIy NDk5Njg=