MusicLM

The Musiclm tool is a model of generating high -quality music based on text descriptions. This model involves a hierarchical task of modeling sequences that generates music in 24 kHz quality, keeping consistency for several minutes. This tool can condition the generated music based on both the text and melody, which allows you to transform screeching and humming melodies, according to the style described in the text signature. This tool can also generate music based on descriptions of images, instruments, genres, levels of experience of musicians, places and eras. In addition, the tool is able to generate a variety of versions of the same text inquiry and the same semantic tokens, providing different interpretations.