Introducing Prof. Sharon Gannot

In recent years, an international project involving eight European teams has been developing a robot with social skills. Funded by the European Union's Horizon2020 program, the project is in its pilot deployment phase at a French geriatric hospital. Prof. Sharon Gannot, heading one of the teams participating in the project, explains: “The robot is intended to be deployed in hospitals as an aid and help reduce staff workload. This means it must be able to interact with people who come for checkups at the hospital and with their escorts. Accordingly, it needs to be able not only to see and hear but also to understand what is being said to it, to have a dialogue, and to identify emotions. If, for example, the robot detects that its interlocutor is repelled by it, not an unlikely prospect, given that older people can be technology-shy, it should know how to apologize and end the conversation. My team is responsible for all aspects of the robot's hearing: its ability to tell the designated speaker from background noises and other speakers. The better the quality of the speech signal recorded by the robot, the more we can assist the robot's understanding ability – which is the responsibility of a different team.”

Prof. Gannot specializes in speech signal processing. He completed his undergraduate degree in Electrical Engineering at the Technion with distinction before his army service (“Atuda” program), then served in the IDF for seven years (and many more in reserve duty), discharged as a Major. For one of his achievements during his military service, he received the Head of Intelligence Branch, Creativity Award in memory of Colonel Uzi Yairi. From there, he went on to complete an M.Sc. degree (with distinction) and a Ph.D. at Tel Aviv University, followed by postdoc research in Belgium at Katholieke Universiteit, Leuven (KULeuven) and another two years at the Technion. “I decided to specialize in the field of signal processing when I started my M.Sc. studies,” he says. “I found the field fascinating, as it combines mathematical approaches with engineering applications. This is a mathematical field, and mathematics has always attracted me. In fact, signal processing is an applied field of mathematics with a wide range of applications in academia and industry. After graduating, my students landed research and development roles in all major companies as algorithm developers and research and development team leaders. Some work for companies like CEVA, General Motors, Apple, Intel, and many other leading Hi-Tech companies. Four of my former students are now faculty members at academic institutions: at the University of Erlangen-Nuremberg in Germany, the Technion, the Technological Institute in Holon, and Tel Aviv University.”

Shortly after its establishment, Prof. Gannot joined the Faculty of Engineering in 2003. “I was a co-founder of the Electrical Engineering program and one of the first to join the faculty. The founding Dean, Prof. George Moschytz, initially recruited Prof. Amir Leshem, another signal processing expert, and a year later, Prof. Ephi Zehavi, Prof. Zeev Zalevsky, Prof. Arie Weiss, and me,” he recalled. “I'm happy to say I'm pleased with my achievements here. I have had over 300 papers published in books, leading journals, and relevant conferences and won 14 outstanding paper awards at conferences or journals. I received a Fellow rank, the highest rank awarded by the IEEE, the leading professional association in the field of electrical engineering and actually the world’s largest professional organization. In 2022, I earned an award from the European Signal Processing Association (EURASIP), whose members are leading worldwide researchers. This annual award honors one researcher, the head of a research group that has made outstanding scientific achievements in the field of signal processing. In addition, I serve in leading roles in the research community, as an associate editor of several important journals and as the head of several important committees in the field.”

Prof. Gannot's research focuses on various aspects of speech signal enhancement, such as noise reduction, speaker separation, extraction of a desired speaker from a group of speakers, echo cancellation, and speaker localization tracking. In his research, he develops multi-channel and distributed algorithms, optimal statistical algorithms, and machine learning methods, specifically deep learning methods for speech processing. “One of my projects, in collaboration with Prof. Elana Zion-Golumbic from the Multidisciplinary Brain Research Center at Bar-Ilan University and Prof. Jacob Goldberger from our faculty, deals with developing algorithms for directing hearing aids by recording brain activity, specifically electroencephalogram (EEG)” says Prof. Gannot. “A hearing aid is supposed to help the user focus on the attended speaker by extracting its voice from all other speakers and noise. In this context, the question that concerns us is how the hearing aid knows to whom it must listen. To this end, we developed an algorithm that uses recorded audio and EEG, an electrical recording of brain activity, to identify which user the hearing aid wearer is listening to.”

A decade ago, Prof. Gannot established the advanced degree study program in Information Processing and Data Science at the Faculty of Engineering, which attracts many outstanding research students. Additionally, five years ago, Prof. Gannot and Prof. Yoel Greenberg from the Department of Music at Bar-Ilan University established the combined track of Electrical Engineering and Music, also focusing on signal processing and machine learning aspects. In recent years, Prof. Gannot has begun developing a bachelor's degree in Data Engineering study program to be launched in the upcoming academic year. "We decided to launch this program because Data Engineering is in high demand across the industry and academia, and there’s a shortage of specific training in this field. The global data revolution has generated a great need for engineers to develop machine learning algorithms. In planning the program, I worked together with Prof. Ran Gelles from the Computer Engineering Program. We looked into other programs worldwide and constructed the program according to the market's needs as we see them. Several faculty members have been recruited to the discipline in recent years: Dr. Ethan Fetaya, Dr. Ofir Lindenbaum, Dr. Tom Tirer, and Dr. Michal Yemini, who joined the more senior faculty members. We hope to recruit additional faculty members in the relevant fields in the coming years. In this context, the prestigious IEEE Signal Processing Magazine will shortly publish an article I co-authored with colleagues from Europe, the U.S., and Southeast Asia, exploring principles of designing modern academic data engineering curricula with an emphasis on machine learning and signal processing.”