Google releases generative AI music tool MusicFX for public use

The new tool marks the democratization of music production

Days after the release of its note-taking tool NotebookLM, Google unveiled another of its groundbreaking AI toys – MusicFX. As you’d probably guessed, the newest platform allows users to generate music using artificial intelligence. All you need is a Google account and a terse description of the melody you’re looking for.

The tool comes from AI Test Kitchen, the platform Google launched last year for rotating experimental AI technology tryouts. First available as a mobile app, it was moved to the web-only format in August. The models AI Test Kitchen previously tried out included generative language models such as the LaMDA chatbot.

In Google’s extensive list of AI experiments, MusicFX is the latest version of the music generator. In May, AI Test Kitchen released MusicLM that offered users to “Describe any musical ideas and hear it come to life with Al.” To make it work, you would write instructions for the music, however vague or detailed, from the mood of the melody to the specific instruments you want to hear. MusicFX is ultimately what came out of that for public use.

In a simple interface, users can write prompts such as “Soulful jazz for wine-sipping event” or “Tango with jingle bells for Santa’s sleigh ride.” The generated music lasts from 30 to 70 seconds and can loop back to the beginning. You can download and share the result – however, it will not pass as your own creation. The end product includes a unique digital watermark from SynthID, developed by Google subsidiary DeepMind.

Google noted that the data collected during the music-generation process is stored anonymously and not linked to the user’s account. The developers were also careful to avoid copyright issues: MusicFX refuses to include specific musicians’ voices and songs.

However, Google does have another platform for users to generate music with the AI-made voice and musical style of several artists. This tool was built on DeepMind’s Lyria model and launched last month under the name Dream Track. Google called it an experiment “designed to help deepen connections between artists, creators, and fans through music creation.” It’s created for use in YouTube Shorts. Among the artists whose vocals can be used are Sia, Demi Lovato, T-Pain, and John Legend.

In parallel with Dream Track, Google announced the development of its music AI tools. These, it said, will help artists, songwriters, and producers use AI responsibly to boost the creative process on YouTube.

“Imagine singing a melody to create a horn line, transforming chords from a MIDI keyboard into a realistic vocal choir, or adding an instrumental accompaniment to a vocal track,” the company wrote in a mid-November blog.

Indeed, moving on from the textual and image-generating applications of AI, the new music tools seem like a natural step into the machine-made abyss. While AI is no Mozart and has no soul that could cry to the beauty of music, its availability signifies the democratization of music production and makes it easier to create - for anyone.

Image: MusicFX