MusicLM, an experimental AI technology developed by Google, can convert written descriptions into musical compositions. MusicLM is a tool inside the AI Test Kitchen app (web, Android, iOS) that enables users to write down in a prompt and have the tool generate multiple versions of the song based on the input. Users can modify their MusicLM-generated creations by specifying instrument types like “electronic” or “classical” and by indicating the “vibe, mood, or emotion” they’re going for.

Conditional music creation is modeled as a hierarchical sequence-to-sequence modeling task in MusicLM, and the resulting music maintains a relentless 24 kHz sampling rate over many minutes. The outcomes of the studies exhibit that MusicLM is superior to competing systems when it comes to audio quality and accuracy within the written description. Researchers at Google show that MusicLM may be trained on the text and a melody, adapting whistled and hummed melodies to match the style described in a text caption. MusicCaps, a dataset including 5.5k music-text combos with wealthy text descriptions produced by human experts, has been made freely available by Google researchers to facilitate further research.

There are 5,521 musical examples within the MusicCaps dataset, each accompanied by a free-text caption written by a musician and an English aspect list. As an example, “pop, tinny wide hi-hats, mellow piano melody, high pitched female vocal melody, sustained pulsating synth lead” is a listing of characteristics. This can be a low-quality recording. Several phrases describing the music are included within the caption. For instance: “A low-sounding male voice is rapping over fast-paced drums playing a reggaeton beat together with a bass.” The accompanying music appears like it’s being played on a guitar. Some are chuckling off in the space. One could hear this tune in a bar. Only the music itself, not any metadata just like the artist’s name, is discussed within the text. AudioSet incorporates 2,858 evaluations and a pair of,663 training examples, each lasting 10 seconds.

Different audio/music generations may be seen here for instance.

In a January research paper, Google previewed MusicLM but specified it had “no immediate plans” to distribute the software. MusicLM, the tactic described within the article, presents quite a few ethical concerns, corresponding to incorporating copyrighted material from training data into the created songs, because the paper’s authors identified. Google has been doing workshops with musicians to “see how [the] technology can empower the creative process.” Possible result? MusicLM, implemented within the AI Test Kitchen, doesn’t produce songs with particular artists or vocals. Take that for what it’s price. The larger problems with generative music don’t have an easy solution.

Popular recently are amateur tracks that employ generative AI to provide recognizable sounds convincing enough to be passed off as real. The music industry has been desperate to alert their streaming partners, citing mental property issues, once they discover recent songs.

Try the Paper, Project, and Dataset. Don’t forget to affix our 21k+ ML SubReddit, Discord Channel, and Email Newsletter, where we share the newest AI research news, cool AI projects, and more. If you could have any questions regarding the above article or if we missed anything, be at liberty to email us at Asif@marktechpost.com

🚀 Check Out 100’s AI Tools in AI Tools Club