Neural Network-based Synthesis for Digital Sound Processing

The allure of synthesizers built upon analog circuits lies in their individuality and idiosyncratic traits, which have captivated enthusiasts since the 1960s. In contrast, the digital age ushered in sound modules that generate a clean and almost predictable sound, setting them apart from their analog counterparts. Venturing into the expansive realm of neural networks, this research delves into the field of machine learning and the significant attention it has garnered in recent years. The widespread application of these technologies now permeates our daily lives. Exploring the intersection of creativity and machine learning, researchers have investigated how these same techniques can be harnessed for expressive purposes. In the domain of digital sound processing, this paper serves as a proof of concept, demonstrating how machine learning methods can synthesize sound and infuse it with an analog essence.