Equivalence (2022/2023): From Speech To Space

Cross-Modal,
Visualization,
Interactive Installation

"Equivalence" a microphone-based installation that takes speech from audiences as input and transforms the signal into visual structures in 3D space based on language processing. This is a colaborative work with Dan C. Baciu and George Legrady. This interactive installation is an artistic work that explores cross-modal interaction’s potential in producing images. The intent is to investigate to what degree the syntactic structure of language can be a means by which to build a stream of varying emergent visual forms.

The project is a contemporary updated version of “Equivalents II” realized in 1992 by Prof. George Legrady , a pioneer new media art text generating visualization installation featured in numerous museums such as the Fotomuseum Winterthur, Switzerland, the Ansel Adams Center, San Francisco, the International Center of Photography, NYC, and other venues.

The microphone captures surrounding sound input and performs real-time Speech Recognition and Audio Analysis. There are three layers of processing: Word-Level (to understand each word), Sentence-Level (to understand each sentence), Document-Level (to understand all the sentences being recognized), representing how languages are interpretated.

The computed language features are then mapped into three-dimensional space and generate architectual-like strcutures by controlling each structure's transformation and rendering features.
The program is built with Python (3.9). For geometry rendering and visual effect manipulation, we use Panda3D (1.10.12).

Information

George Legrady: Concept & Creative Direction
Dan Costa Baciu: Natural Language Processing, Architecture Design
Yixuan Li: Machine Learning based Software Development