Sihwa Park


Selected Works
YouTube Mirror
Uncertain Facing
Ballet Mécanique
Brand Logo Sonification
Structured Improvisation X

Letter Frequency Visualization
Generative Audiovisual Study
Ring of Quartets
Hans Zimmer’s OST vs Film
︎︎︎Past Works (2008-2015)

Recent News

- Apr 7, 2024: YouTube Mirror has been accepted for CVPR AI Art Gallery!
- Oct 24, 2023: Appointed a Connected Minds Research Enhanced Hire position with a $100,000 startup fund.
- Sep 1, 2023: YouTube Mirror was presented at AIMC 2023, Brighton, UK.
- Jun 12, 2023: YouTube Mirror has been accepted for AIMC 2023 as both a paper and an installation!
- May 31, 2023: YouTube Mirror was exhibited at NIME 2023, Mexico City, Mexico.
©2023 Sihwa Park

Sihwa Park

YouTube Mirror

Date: May 2022
Categories: AI Art, Data Art, Audiovisual Installation, Generative

YouTube Mirror is an interactive, audiovisual AI installation that generates images and sounds in response to the images of the audience captured through a camera in real time, in the form of an interactive mirror. YouTube Mirror uses a cross-modal generative machine model that was trained in an unsupervised way to learn the association between images and sounds obtained from the YouTube videos I watched.

YouTube Mirror employs a concept of interactive mirror in digital media arts, as a way to represent how machine vision perceives the world. Since the machine represents the world solely based on its learned audio-visual relationship, the audience can see themselves through the lens of the machine’s vision with the machine-generated images and sounds.

YouTube Mirror is an artistic attempt to simulate my unconscious, implicit understanding of audio-visual relationships that can be found in and limited by the videos. YouTube Mirror also attempts to represent the possibility of the machine’s bias inherent in a small-size video dataset which is possibly affected by video recommendation algorithms on YouTube, in an abstract and artistic way.

YouTube Mirror at NIME 2023, Mexico City, Mexico

This version of the installation accompanied a video grid that loops 300 YouTube clips randomly selected among the whole 6,091 video data in order to show what data was used for training the model.

YouTube Mirror at AIMC 2023, Brighton, United Kingdom

Related Publication
Sihwa Park, “YouTube Mirror: An Interactive Audiovisual Installation based on Cross-Modal Generative Modeling”, In Proceedings of the International Conference on AI and Musical Creativity (AIMC), Brighton, United Kingdom, 2023

Related Exhibitions
Jun 2024, “YouTube Mirror” at the IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR) AI Art Gallery 2024 (Online)
Aug 2023, “YouTube Mirror” at the International Conference on AI and Musical Creativity (AIMC), Attenborough Centre for the Creative Arts, Brighton, United Kingdom
May 2023, “YouTube Mirror” at the International Conference on New Interfaces for Musical Expression (NIME), Centro de Cultura Digital, Mexico City, Mexico
May 2022, “YouTube Mirror” at the MAT 2023 End of Year Show: SYMADES, California NanoSystems Institute, UCSB, Santa Barbara, USA