These works stem from my recordings of transnational contemporary music, weaving together traditional instruments: Armenian, Indian, and Serbian idioms, with contemporary ensemble textures. Leveraging Stability AI’s advanced audio diffusion models, such as Stable Audio, and hybrid transformation tools, I crafted new soundscapes that transcend mere remixing.  AI employs diffusion processes, adding and then denoising noise in data using neural networks like transformers, to reinterpret sonic material. This process converts audio into spectrograms, infuses them with controlled randomness, and regenerates them into unexpected forms, guided by my input or text prompts.

The AI acts as a “foreign ear,” reinterpreting cultural memory and sonic identity through its algorithmic lens. Trained on vast, diverse datasets (e.g., global sound archives), it detects patterns beyond my intent, transforming folk melodies into polymorphic voices or Indian rhythmic structures into expansive acoustic spaces. I intentionally introduce “glitches” by manipulating latent representations or adding noise, reflecting human fragility and opening new creative territories.

​

​