About MusicLM

MusicLM is a text-based AI model that revolutionizes the world of music generation and composition. Developed using the state-of-the-art artificial intelligence architecture, MusicLM can create original music across various genres and styles based on text prompts. The model produces rich, high-fidelity melodies from simple text descriptions, redefining conditional music generation through a sophisticated hierarchical sequence-to-sequence modeling process. The underyling dataset consists of 5.5k curated music-text pairs. Each pair includes a carefully crafted text description provided by human experts, offering a wealth of opportunities for researchers and music enthusiasts alike.

In a Nutshell

Demo

Click below to try a demo of MusicLM:

MusicLM Demo

Authors

MusicLM was built by a Google Research team involving Andrea Agostinelli, Timo I. Denk, Zalán Borsos, Jesse Engel, Mauro Verzetti, Antoine Caillon, Qingqing Huang, Aren Jansen, Adam Roberts, Marco Tagliasacchi, Matt Sharifi, Neil Zeghidour, Christian Frank.

Get Started

Use the links below to browse the MusicLM GitHub page, the research paper as well as the underlying dataset.

MusicLM on GitHub MusicLM Research Paper MusicLM Dataset