AWS DeepComposer part 2 generating music

Transcript

Alright, we're back with AWS Deep Composer. In the previous video, I showed you the keyboard itself. Now, let's talk about the service. You should go to aws.amazon.com/deepcomposer, where you can sign up for the preview and get notified when the keyboard becomes available. Once you're accepted into the preview, you can open the console and see the screen. So, let's compose music and dive straight into the subject. Here, we see a virtual keyboard, so you don't need a physical keyboard to play. Just make sure you've enabled hotkeys in the settings because this will let you use your computer keyboard to play. You can play multiple notes, which you won't be able to do if hotkeys are not enabled. Of course, if you have a physical keyboard plugged in, like me, you can play as well, and this doesn't require any setup. I just plugged the keyboard in, and it worked out of the box. That's pretty cool. We also have a bunch of songs to get you started. Let's check this one out. These are pre-recorded songs you can use to generate music. To actually generate music, you need models. If you go to models, you can create models, meaning you can use pre-trained models. These are the ones we have right now: jazz, pop, rock, classical, and the Jonathan Coulton model. Jonathan Coulton is a musician who contributes to the Deep Composer project. Let's grab the pop model and give it a name. Create it. I'll talk about training your own model in a future video. This is a much more advanced project, and I just want to give you an overview here. So, we have models, and I've already added the other ones. Now, let's try to generate music. We'll select "Ode to Joy" again and use the pop model. Just click on Generate Composition. In a few seconds, this generative model adds extra instruments to the song. I believe they're called compositions, which sounds like a weird word, so I'm just going to say instruments. It's picking actual instruments for you automatically, and you can try different ones. Let's try this version first.