Music / Simulation / Casual ⚡ Runtime LLM
Blob Opera
Created by Google Arts & Culture & David Li Direct Web Game
Conduct four charming musical blobs that harmonize operatic songs in real time using a machine learning vocal model.
Touch supported Controls & tipsAI Architecture & Technology
Architecture & Workflow Breakdown
Blob Opera utilizes a machine learning model trained on sixteen hours of singing by four professional opera vocalists. As you drag the four animated blobs to modulate pitch and vowel sounds, the neural network synthesizes operatic harmony in real time without playback of pre-recorded audio tracks.
About Blob Opera
Blob Opera is a whimsical and technically dazzling musical experiment developed by David Li in collaboration with Google Arts & Culture. The game features four squishy, colorful operatic blobs representing the classic vocal ranges: Bass, Tenor, Mezzo-Soprano, and Soprano. Rather than playing back pre-recorded sound samples, Blob Opera runs a machine learning vocal synthesis model trained on sixteen hours of singing from four world-class opera vocalists. By clicking and dragging any blob vertically, you manipulate its pitch; dragging horizontally morphs its vowel resonance from soft 'ooohs' to piercing 'aaahs'. The magic happens in the real-time AI harmonization: as you steer one blob melody, the other three instantly adjust their pitches and formants to harmonize in pitch-perfect counterpoint. You can record your own operatic compositions, toggle festive seasonal backdrops, or sit back and let the blobs perform classic carols.
Controls & How to Play
Controls follow standard web interaction: mouse, touch.
📱 Mobile touch controls supported.



