Real-time audio synthesis in the browser is architecturally fundamentally different from the traditional approach of streaming fixed MP3s: the former's music library is a set of generative rules, while the latter's is a finite collection of audio assets.