This project, conducted during our senior year engineering capstone, aimed to explore automated music generation through programmed music theory and pattern recognition. Divided into three key segments: Musical Algorithms, Pattern Recognition and Extraction, and Music Composition Generator, we sought to develop a system capable of autonomously crafting musically coherent compositions, with each segment building the foundation for the next.
Music theory rules were encoded first so later pattern extraction had a structured base to operate on.
Pattern recognition stages focused on identifying recurring note relationships that could be reused compositionally.
The generation pipeline explored how algorithmic structure could produce coherent musical output instead of random sequences.
In this initial phase, a deep-dive research was conducted to understand the algorithmic essence in music theory. Various code models mirroring this algorithmic nature were studied, analyzing music and setting the groundwork for the creation of new melodies, such as Chords and Triads, Cadences, Musical Scales, Rythm and Time Signatures.
Moving on to the next section, we chose sheet music instead of sound files, in line with our main emphasis on music theory. We encoded the sheet music into '.ABC' format, and fed hundreds of compositions into our system, which aided in identifying recurring patterns. This data served as a fundamental resource for grasping common musical structures.
The final stage of our project was the merging of the identified musical patterns. Utilizing our earlier developed musical algorithm models, we aimed to replicate the complex process of music composition. The integration of these patterns through our algorithms led to the creation of new, coherent songs, thus fulfilling our objective of automated music generation.
This project was conducted during senior-year engineering capstone work and focused on automated music generation through programmed music theory, pattern recognition, and composition automation.
The system was structured in stages so that music-theory modeling supported pattern extraction, and the extracted structures then fed the composition generator.
In this initial phase, a deep-dive research was conducted to understand the algorithmic essence in music theory. Various code models mirroring this algorithmic nature were studied, analyzing music and setting the groundwork for the creation of new melodies, such as Chords and Triads, Cadences, Musical Scales, Rythm and Time Signatures.
Moving on to the next section, we chose sheet music instead of sound files, in line with our main emphasis on music theory. We encoded the sheet music into '.ABC' format, and fed hundreds of compositions into our system, which aided in identifying recurring patterns. This data served as a fundamental resource for grasping common musical structures.
The final stage of our project was the merging of the identified musical patterns. Utilizing our earlier developed musical algorithm models, we aimed to replicate the complex process of music composition. The integration of these patterns through our algorithms led to the creation of new, coherent songs, thus fulfilling our objective of automated music generation.
We were able to generate unique music, which notably caught the attention of a CBC News reporter during our presentation day at the 2022 UNB Engineering Symposium. The event, which was a significant platform to showcase our project, turned more exhilarating as the reporter, amidst the attendees, took a keen interest in our work. The positive feedback we received from everyone present not only bolstered our confidence but also highlighted the impact and the potential our project holds in the intriguing intersection of music and technology.
The project marked a fresh and challenging venture into the intersection of music and technology, reigniting my early acquaintance with music theory while significantly testing our programming skills and knowledge acquired from courses. It propelled us into a continuous learning journey, and opened new potential future explorations in this domain.