rryam/musaverakit

MusaveraKit is a Swift package that makes it simpler to use Apple's Music Understanding framework in iOS, macOS, tvOS, watchOS, and visionOS apps. Apple introduced Music Understanding at WWDC26 as a way to analyze audio files directly on the device, without sending audio to any server. MusaveraKit wraps that framework with shorter, cleaner code so you spend less time on session setup and repetition. The package works by taking an audio file your app already has access to and running an analysis on it. You get back information about the music: beats per minute, musical key, loudness, perceived pace, the number of sections in the song, and whether vocals, drums, bass, or other instruments are active at different points. All of that processing runs on-device, so no audio ever leaves the user's device. If you only need one aspect of a track, the package has focused helper functions for rhythm, key, loudness, pace, structure, and instrument activity individually. There is also a streaming mode for analyzing live audio as it comes in, for example from a microphone, so you can read loudness values in real time while the rest of the analysis finishes once the audio stream stops. For apps that show music visualizations or playback interfaces, MusaveraKit adds timeline helpers. You can look up the musical key at a specific playback position, find the nearest beat or bar to a timestamp, or check which song section or phrase is active at a given moment. Because this builds on a framework still in beta as of mid-2026, APIs may change as Apple updates the SDK. The package requires Xcode 27, Swift 6.4, and Apple platform operating systems version 27 or later. It is distributed as a Swift Package and is available under the MIT License.