Smart audio devices (Google's assistant, Siri, Amazon's Alexa...) are today extremely popular. While machine learning and artificial intelligence have traditionally focused on speech, images or music, the general understanding of “sounds” has not been a domain of interest until recently. The seminar will present at a very basic level the traditional approaches for sound recognition based on hand-crafted features, and will discuss how recent approaches based on deep neural networks have taken the lead in the past few years. The main open problems in this field will be presented, demonstrating that there is still much work to be done in this context.

Maximo Cobos is an Associate Professor with the Engineering School of the Universitat de Valencia (Spain). He received the Ph.D. degree in telecommunications from the Universitat Politècnica de València in 2009, and worked for the Institute of Telecommunications and Multimedia Applications after being awarded by the "Campus of Excellence" post-doctoral fellowship. His research focuses on the area of signal processing and machine learning for audio and multimedia, with application to source localization, separation, identification, among others. He is a Senior Member of the IEEE, a full member of the Acoustical Society of America (ASA) and a member of the Audio Signal Processing Technical Committee of the European Acoustics Association (EAA).