You can create music in seconds with Google’s AI

While last week we mentioned that we’ll have to wait a bit longer to try Google‘s new AI chatbot, the company has released another AI that will definitely catch your attention. The new experimental tool, Instrument Playground, allows users to create 20-second musical tracks using up to 100 different musical instruments from around the world, all thanks to the MusicLM AI.

Bard ACCESS

MusicLM can match the sound of the instrument with adjectives like “cheerful” or “romantic” to create a specific sound. For more experienced musicians, the tool even features a sequencer that allows the creation of layers and loops with up to four different generated instruments, each with its unique sound.

Google states that Instrument Playground’s starting point was to explore a MusicLM-based interface that would inspire creativity and the discovery of instruments from around the world.

Instrument Playground is an expansion of Google’s MusicLM project, a generative artificial intelligence model capable of creating high-fidelity music from text inputs. Google introduced MusicLM in early 2023, stating that it was trained on a dataset of 280,000 hours of music to produce songs from text messages.

Researchers claim the model has been developed with the ability to transform a sequence of written descriptions into a musical story, building upon existing melodies that can be whistled, hummed, sung, or played on an instrument.

To enable users to create music from scratch with up to 100 instruments, Google had to train MusicLM with hundreds of hours of sounds from each of these instruments. Whether it’s a viola, an ocarina, or a saxophone, the algorithm has received an extensive amount of performance hours for these instruments to accurately recreate their sounds.