FROM RADIO TO VIDEO
Interactive Installation, AI Systems, 
Real-Time Media, Experimental Design.
This project explores real-time translation between sound and image using open-source AI and node-based programming.

We hacked a radio and transformed its live audio signal into an AI-generated video stream. The system captures what is being broadcast, transcribes it using Whisper, and processes the text through a language model to extract five key words. These keywords are then used as prompts to generate visuals in real time through TouchDesigner’s Stream Diffusion operator.

A physical button added to the radio allows interaction with the system by altering the audio noise. This directly affects the level of abstraction in the generated visuals, creating a dynamic relationship between sound, language, and image.

In collaboration with Malik Ahmed and Jose Pardo.