C-BRAHMS research focuses on three computational musicology areas: 1) music pattern matching and detection, 2) automated music generation with the help of human computer interaction, and 3) score following with the help of the first area of this project, optical music recognition and audio transcription.
WORK PACKAGE 1: Symbolic music pattern matching and detection

This work package investigates the classical music pattern matching and detection problem. Here, we use the point set representation, notable for its ability to apply various musically meaningful invariances in search algorithms. While this representation offers advantages, it also introduces computational complexity and generates many musically insignificant matches due to applied invariances. In this work package, we also develop machine learning methods that automatically prune musically insignificant mass occurrences that overshadow the few meaningful instances.

WORK PACKAGE 2: Music generation , human computer interaction

This work package concentrates on automatic music generation using AI and machine learning advancements for music generation with the humanin the loop approach. The considered techniques include, for instance, Short Long Term Memory (SLTM) and Generative Adversarial Networks (GANs). The global research of this work package is evolving rapidly due to recent advancements in AI and generative language models.

WORK PACKAGE 3: Score following and audio transcription

This work package focuses on score following that entails real time tracking of a live music performance against a pre written score or notation. Such a method facilitates the synchronization of live musicians with automated systems, such as digital accompaniments, interactive music systems, or visual displays, by precisely identifying the current position within the score. A promising new approach to this end appears in implementing synchronization using fault tolerant and efficient algorithms developed in work package 1 in the form of point set music data. The music scores are converted into this format using optical music recognition, and the audio using transcription algorithms.