20 Years of Musical Genre Classification - Past, Present, and Future

In 2000, compact discs were the dominant mode of music distribution and the use of digital compressed audio files for playback using computers was just beginning. Most of the research on analyzing music was done with symbolic representations rather than audio signals. The presenter, Dr. George Tzanetakis, is fascinated by pattern recognition, computer vision, and speech recognition. His curiosity on the topic led him to research whether similar approaches could be followed to analyze audio recordings of music. His most influential and pioneering publication ?Musical Genre Classification of Audio Signals? was the product of these investigations and heavily influenced the development of research field of Music Information Retrieval (MIR).

In this webinar, Dr. Tzanetakis will discuss the main ideas introduced in the paper, provide some historical context about their development, and trace their impact over the last 20 years. By telling this story, he hopes to encourage graduate students and young researchers to think about how to conduct research that has long term impact.