Imagine a karaoke machine that uses AI to blend two of your songs into an entirely new musical creation. That’s the idea behind ‘DeepKaraoke’, the outcome of a project exploring machine learning as a design tool.
The experience starts with a physical cube with album art on each face. Using computer vision, two selected songs are detected by a camera. Then, the AI generates new lyrics inspired by both songs, creating a completely original mashup of music and words.
The new lyrics are displayed over an instrumental of the first song, with a live music video also generated using the video feed from the camera combined with random visuals.
Select two songs using the song cube.
Computer Vision detects the album art.
The ML model generates original song lyrics based on the two selected songs.
A live music video is generated using the video feed.
Sing it out!