Amazon’s DeepComposer Marks The Beginning Of The Robots Taking Over Music

Nothing like starting the new year on a positive note : ) Amazon recently announced a new initiative to bring machine learning to a wider audience of users, and one way is with something that it’s calling AWS DeepComposer, “the world’s first machine learning-enabled musical keyboard for developers.”

DeepComposer basically consists of a keyboard that connects to the AWS (Amazon Web Services) cloud and uses something called Generative AI to totally arrange a song based on just a melody that’s played into the instrument. This includes all of the supporting parts like drums, bass, guitar, keys, strings, and anything else you can think of. It can shape your melody in a number of broad musical genres, including pop, rock, jazz, classical, or one that you choose to construct.

Generative AI is different from most machine learning processes in that it actually uses two different networks to accomplish a task. A Generator network creates the music (in this case) based on the training that you give it, and a Discriminator network provides feedback on what to improve, so both networks train each other.

Once your composition is completed you can then publish it to your SoundCloud account.

The DeepComposer keyboard will cost $99 and be connected to a free tier that gives you access to the machine learning cloud for 12 months. The full account pricing is based on a usage model. It usually takes about 8 hours for the process to be trained based on your input, and each training hour costs just $1.26. After that the execution phase costs $2.14 per hour, but the process is so fast that it takes less than a minute to complete the task. That means that the costs of your first full composition should be around $10, and then less than a $1 after that!

DeepComposer is not quite ready for release yet but it doesn’t appear to be too far off. You can sign up for it here and also find out much more about it, or see how it works via the video below.

---