Developers: | |
Date of the premiere of the system: | May 2023 |
Branches: | Internet services |
2023: Product Announcement
On May 10, 2023, the company Google introduced the open source MusicLM platform based on. artificial intelligence This service is designed to create musical works by text description.
Access to MusicLM is provided based on the AI Test Kitchen application through a browser, as well as on mobile devices running Android and iOS. The user can enter a natural language request, such as "sentimental jazz" or "create techno sound." In response, AI will generate two versions of the product, from which you can choose the most suitable one. In queries, you can specify the genre of the composition, emotional coloring and other parameters.
However, there are a number of restrictions. In particular, MusicLM does not generate music with vocals or works in the style of specific artists. This is due to possible claims from copyright holders. Despite the fact that this generative AI was trained on hundreds of thousands of hours of music recordings, the results of its work often leave much to be desired. The created works may largely not correspond to the theme line set by the user. In addition, sometimes a heap of sounds appears, and the works themselves unexpectedly break off. MusicLM functionality also suffers due to artificial constraints on the query side.
In general, as noted by observers, MusicLM works much better than previously presented systems of this type. Melodies generated by MusicLM contain fewer artifacts. It is assumed that music created using this neural network will be able to find application in various fields, including soundtracks for video games, background accompaniment for podcasts, certain video broadcasts, etc.[1][2]