Program Overview

Master in Sound and Music Computing

The Master in Sound and Music Computing is a program that focuses on the study and application of sound and music computing techniques.

Overview

The program provides students with a comprehensive understanding of the theoretical foundations and principles of sound and music computing, including symbolic music generation techniques, deep learning-based methods, and raw audio generation techniques.

Academic Program

The academic program consists of various courses, including:

• Generative Algorithms for Sound and Music
  • Instructor: Valerio Velardo, Lonce Wyse
  • Credits: 5 ECTS
  • The course focuses on exploring, implementing, and critically evaluating symbolic music generation techniques, including traditional approaches and deep learning-based methods.
  • The course also covers raw audio generation techniques using deep learning-based methods.

Learning Objectives

The learning objectives of the course are divided into two parts:

Part One

• Understand the theoretical foundations and principles of symbolic generative music techniques.
• Analyze the strengths and limitations of both traditional and machine-learning based symbolic generative music techniques.
• Code generative music systems from scratch.
• Connect theory to practice by exploring how generative music techniques are applied in research, industry, and artistic projects.
• Combine, adapt, and implement generative techniques to design symbolic music systems tailored to specific creative tasks and musical objectives.
• Recreate and implement generative systems described in research papers.
• Run inference and fine-tune pre-trained symbolic models.
• Leverage cloud platforms to run generative music models on remote servers with GPU acceleration.
• Discuss the skills, knowledge, and steps required to become a generative music AI engineer.

Part Two

• Develop an understanding of the deep learning techniques used for generative audio.
• Develop/deepen the ability to read and understand research papers in the field.
• Understand not just how to use, but how to build or change networks to achieve goals.

Concepts

The course covers various concepts, including:

• Other “subsymbolic” audio representations frame-based codecs
• Conditioning
• Managing long time dependencies
• Latent spaces for creative interaction
• Managing computational overhead
• Playability
• Offline vs online / Composition vs performance / Text vs RT Interaction

Architectures

The course covers various architectures, including:

• Convolutional (e.g. Wavenet)
• GAN (e.g. Sound Model Factory)
• VAE (e.g. Rave)
• DDSP (e.g. DDSP)
• Transformers (Vampnet)

Pre-requisites

The pre-requisites for the course are:

• Intermediate proficiency in Python programming.
• Basic understanding of linear algebra.
• Basic knowledge of TensorFlow/Keras and PyTorch will be helpful for deep learning techniques.

Teaching Approach

The teaching approach of the course is based on:

• Learning by doing
• Fostering proactivity and independent learning The course will cover both the theoretical and implementation aspects of symbolic generative music systems. Theory classes will focus on advanced aspects of the techniques, while practical classes will center on code implementation of the assignments.