Visualization of Musical Tension in Salsa
Discipline: Computer Sciences and Information Management
Subcategory: Computer Science & Information Systems
Session: 4
Room: Farragut North
Gabriel A. Santiago Plaza - University of Puerto Rico Río Piedras Campus
Co-Author(s): Dr. Rafael A. Arce Nazario, University of Puerto Rico Río Piedras Campus, San Juan, PRDr. Rémi Mégret, University of Puerto Rico Río Piedras Campus, San Juan, PRDr. Randall E. Cone, Salisbury University, Salisbury, MD
Musical tension is essential for composition and human perception of a score [3]. People perceive different emotions when hearing tension in music and experience relaxation as it resolves [4], playing a pivotal role in the richness of a score [1]. Current methods that use music generating artificial intelligence (AI) fall short in creating quality musical content within different styles since their human-computer interaction is limited and/or they generate monotone melodies [2]. Because of this, these methods become inadequate when used in Salsa music since they lack the ability to produce art that meet the expectations of custom and musically-enriched pieces that represent part of our Caribbean Culture. For this reason, we set a goal to create an AI-assisted Music Composition Tool that takes in account user-defined tension-building patterns to generate and provide possible melodies to be added into the composed score.As the first stepping stone in that direction, we propose an interactive visualization tool that displays a structural representation of tension build-up and release within single-track polyphonic MIDI files. We do this by implementing music theory based algorithms that detect when and how the tension is building up. This allows users to recognize core musical patterns that are used in Latin genres to reach certain tension, enabling the user to extract useful features in musical structure and behavior. Our preliminary results include a system that detects and visualizes harmony in Bach’s preludes and fugues using graph traversing algorithms that take in account the temporal relationship in each voice. The proposed visualizations rely on new numerical data structures that we will generalize in the future into hierarchical music structures that resemble musical tension. These structures will be used as key elements in Music Generating Models to give musicians the power to control high levels in harmony and visualize the emotional expressiveness of their music. We believe that visualization tools can aid in envisioning the emotions a score is emitting, which can facilitate the creative process of Salsa composers while also giving a sense of direction to Automatic Music Generating Models.References:[1] Douek, J. (2013). Music and emotion—a composer’s perspective. Frontiers in Systems Neuroscience, 7. https://doi.org/10.3389/fnsys.2013.00082[2] IEEE Xplore Full-Text PDF: (n.d.). https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9966445&tag=1[3] Sun, L., Hu, L., Ren, G., & Yang, Y. (2020). Musical tension associated with violations of hierarchical structure. Frontiers in Human Neuroscience, 14. https://doi.org/10.3389/fnhum.2020.578112[4] Zhang, N., Sun, L., Wu, Q., & Yang, Y. (2022). Tension experience induced by tonal and melodic shift at music phrase boundaries. Scientific Reports, 12(1). https://doi.org/10.1038/s41598-022-11949-4
Funder Acknowledgement(s): I would like to thank PR-LSAMP Bridge to the Doctorate Program, funded by the NSF for supporting this project and helping me reach my goals as a PhD Candidate.
Faculty Advisor: Dr. Rafael A. Arce Nazario, rafael.arce@upr.edu
Role: With the guidance and support of Dr. Randall E. Cone, it helped me on the development of weighted directed graphs that represent possible harmonic movements. I also worked with the graph traversing algorithms, and the visualization algorithms that helped us see the harmonical movements in each voice. The development of the interactive visualization tool that plots structural data that represents tension was possible with the guidance of Dr. Rafael Arce and Dr. Rémi Mégret. It was possible by implementing rules that use intervals in music to detect tension in single-track polyphonic MIDI files.