https://editor.p5js.org/Raven2950/sketches/CjHUvrKx-

Blob Opera - Google Arts & Culture

🎵 Blob Opera

Blob Opera is an interactive music experiment launched in 2020 by Google Arts & Culture in collaboration with artist David Li. Users drag four “Blob” characters on the screen, and the system generates harmonies in real time, allowing anyone to “conduct” a piece that resembles operatic choral music.

1️⃣ What type of machine learning models did the creator use?

🔹 Core Model Type: Sequence Generation Model

Blob Opera uses a deep neural network sequence model, conceptually similar to:

• RNN (Recurrent Neural Network)
• More specifically, LSTM (Long Short-Term Memory)
• Or other improved time-series modeling architectures designed to capture pitch and temporal relationships

Music is fundamentally a time-series problem:

The current note depends on the previous note.

The current harmony depends on what the other voices are singing at the same time.

Therefore, the system requires a neural network capable of modeling temporal dependencies and sequential structure.

🔹 Functional Structure of the Model

The model primarily performs two key tasks:

① Pitch Mapping