The AWS DeepComposer console was launched at the company’s re:Invent conference in Las Vegas yesterday, and can build an accompaniment to a simple melody using artificial intelligence (AI).
The company says the device includes tutorials, sample code, and training data that can be used to get started building generative models, all without having to write a single line of code.
Matt Wood, VP of AI at AWS, explained the AWS DeepComposer is powered by generative AI. He said, “generative AI is a little different from all other machine learning approaches in that it uses two networks which train each other.”
That means combining two models – a generator and a discriminator. Here, the generator model generates the music, whereas the discriminator model tries to provide feedback to the generator to help improve the output.
Wood added, “One way to think about this is that the generator is a lot like an orchestra. It can play music, anything that you want. And the discriminator is a lot like a conductor, it can take music, it can provide its own interpretation and critically it provides feedback to the generator, to the orchestra, on how to play and how to improve.”
AWS has said DeepComposer will be initially available in the US from $99.