MFCC Soundscape
The AI Sampler project is a blend of advanced machine learning techniques and classic audio effects, designed to create unique and evolving soundscapes. Using MaxMSP, the system employs MFCC (Mel-frequency cepstral coefficients) analysis to break down sound files into their fundamental features, and then utilizes Fluid UMAP to organize these sounds into clusters based on their similarity. This creates a dynamic and fluid arrangement of sound slices, which results in a constantly shifting sonic landscape. The AI-driven process unveils hidden connections between different sounds, offering new possibilities for exploration and creativity in sound design.
Alongside these AI-based methods, traditional effects like reverb, filters, and live gain modules are integrated to shape the overall texture of the soundscapes. The biquad~ filter, for instance, is used to fine-tune the frequency spectrum, while the Gigaverb module adds a lush sense of space, enhancing the immersive quality of the audio. Together, these elements work in harmony, blending cutting-edge AI technology with conventional sound manipulation techniques to craft soothing, ambient soundscapes that balance both algorithmic precision and human warmth.
Max/MSP