Tvorba audioreaktivního systému pro živá vystoupení

This bachelor thesis focuses on music visualization using audio-reactive systems. The main goal of the thesis is to design and create a functional prototype of an audio-reactive system for live performances and provide a comprehensive documentation of the process, which can serve as a methodological basis for other creators in this field. The thesis first defines the theoretical context of music visualization and generative art, serving as a basic introduction to the topic, and further describes the technical principles of sound processing and audio-reactivity used in the practical part. Based on a comparative analysis of software tools, TouchDesigner was selected for the practical implementation, primarily due to its unique combination of performance, architecture connecting a robust graphics engine with sound processing, Python integration, and the availability of a non-commercial license. The core of the thesis is the experimental part dedicated to testing sound signal analysis methods and various visualization techniques. The results demonstrated that for the dynamic environment of live concerts, basic analysis methods are insufficient due to the necessity of manual recalibration and susceptibility to errors. Therefore, the external tool ARE (Audio Reactive Engine) was determined as the optimal solution, ensuring reliable synchronization of visuals with music thanks to automatic threshold calibration. The efficiency of GPU-accelerated particle systems (POPs) was verified, enabling the simulation of complex organic structures in real-time while used as the visual core. The main output of the thesis is a functional prototype that autonomously reacts to electronic music in real-time. During its development, it was identified that data preprocessing and mapping are critical factors for the aesthetic quality of the visualization, as they eliminate unwanted digital noise from the raw signal. Another contribution of the thesis is the formulation of a general workflow that organises the creative process into five phases (concept definition, data analysis, core construction, post-processing, and mapping) and serves to minimize risks in generative content creation.