Music Information Retrieval (Lecture + Exercise), Wed 14-18h­­
The digital revolution in music distribution, storage, and consumption is leading to a great demand for automated tools to organize, navigate, retrieve, and analyze large music databases. At the same time, such techniques are also of great importance for computational musicology as a subfield of the "Digital Humanities." In particular, music audio recordings pose challenges for the development of algorithms. As a result, the field of Music Information Retrieval (MIR) has developed into an independent research area at the intersection of different disciplines such as signal processing, information retrieval, machine learning, musicology, and the digital humanities. This course introduces the field of MIR. It teaches fundamentals of music representations (especially audio signals) and music theory concepts. A core focus is on audio signal processing algorithms, especially time-frequency transformations, as well as selected machine learning methods. Furthermore, the lecture gives an overview of MIR tasks (e. g., harmony analysis & chord recognition, beat tracking & tempo estimation, structure analysis, genre & style classification), of which selected applications are considered in depth. Finally, we address the challenges of preparing and annotating large music corpora and the application of MIR algorithms for analyzing such corpora in the context of digital humanities and computational musicology.

Contents: This lecture introduces the research field of Music Information Retrieval (MIR), focussing on the following topics:

• Music representations (graphical, symbolic, audio), basic music theory concepts,
• Audio signal processing (esp. time-frequency transformations, variants of the Fourier transform), selected machine learning techniques
• Overview and in-depth study of individual MIR tasks (e.g., harmony analysis/chord recognition, beat tracking/tempo, structure analysis, genre/style classification)
• Data preparation/annotation and corpus analysis for digital humanities/musicology

Details:

• Lecture slot: Wednesday 14-16 h
• Exercise slot: Wednesday 16-18 h
• Place: TBA (new ZPD building or SE10, physics building)
• SWS: 2 + 2 (Lecture + Exercise)
• Language: English
• Credits: 5 ECTS
• Types of Examination: Oral, German or English

Study programs:

• MSc Computer Science
• MSc xtAI
• MA Digital Humanities

Prerequisites (MSc Computer Science / xtAI):

• Very good mathematical foundations
• Advanced programming skills in Python
• Basic knowledge of deep learning techniques
• Basic knowledge in reading music (treble and bass clef)

Prerequisites (MA Digital Humanities):

• Good mathematical foundations
• Basic programming skills in Python
• Basic understanding of digital editions and corpus analysis
• Basic knowledge of deep learning techniques
• Basic knowledge in reading music (treble and bass clef)
• Special knowledge in music theory is not necessary, but will be taught in the lecture.