Course Description
This course will introduce you to a multidisciplinary field of music information retrieval, that deals with extracting information from audio and symbolic music representations. Would you like to know, how does Youtube find out that a video has a copyrighted audio track? How Spotify knows, what music would you enjoy? How does one create a karaoke version of a song by separating the melody from the rest? In this course, you will find out the magic behind many such techniques for working with musical audio. We will also discuss the properties of the human auditory system and learn, how to extract the most suitable data representations from the digital audio signal. In the second half of the course, we will discuss the application of the state of the art deep learning techniques in music information retrieval. Topics:- Signal processing basics
- Mel-spectrogram, chromagram, audio fingerprinting
- Source separation, applications of triplet networks
- Transformer, CNN in MIR
Course tools
- Python
- Jupyter notebooks
- pandas, librosa, essentia, keras
Prerequisites
- Linear algebra
- Python
- Basics of neural networks