Before Computers: Mozart's Musical Dice Game
The idea of using a systematic process to generate music predates computers by centuries. In 1787, Mozart (or a contemporary using his name) published a 'Musikalisches Wurfelspiel' — a musical dice game. A player would roll dice to select pre-composed musical fragments from a table, then combine them in sequence to create a waltz. Each fragment fit harmonically with the others, so any combination produced a serviceable piece. This is arguably the first algorithmic composition system: a set of rules that generates music without moment-to-moment human input.
The 1950s and 60s: Computers Enter the Picture
The first true computer-generated compositions emerged in the 1950s. Lejaren Hiller and Leonard Isaacson used the ILLIAC I computer at the University of Illinois to generate the Illiac Suite for string quartet (1957). They programmed rules of counterpoint and harmony into the computer and had it choose notes that satisfied the constraints. Around the same time, Iannis Xenakis used mathematical probability theory to compose orchestral works, translating physical models like gas diffusion into musical parameters. These pioneers established the conceptual framework that modern AI music tools build upon.
From Rules to Machine Learning
Many early computer-music systems used explicitly programmed rules, while modern machine-learning systems can learn statistical patterns from examples. Google confirms that Blob Opera uses a second model for harmonization, but it does not publish whether that model learned from scores, audio, rules, or another representation. The broader history helps frame the experiment without filling that gap with speculation.
Where Blob Opera Fits in This History
Blob Opera creates a responsive loop: a person chooses gestures while trained models produce voices and harmony. It is better described as an interactive instrument than as a system that autonomously composes a complete opera. Comparisons with historical chance music or modern generators can be useful, but they are interpretations rather than claims made by the creators.