Programming Voice Interfaces

Chapter 5. A Different Approach Using IoT Core and API.AI

So far we have built a Raspberry Pi–based AVS and created our very first lambda function to support a custom Alexa skill. It’s safe to say, we have accomplished quite a bit in the past few chapters. In this chapter, we are going even further, but with a twist. Rather than further extending our custom Alexa skill, we will bring another platform into the mix to keep things fun and interesting. We are going to build an Alexa-like device using Windows IoT Core running on a Raspberry Pi. On top of all that, we will leverage Google’s API.AI as our cloud-based NLU/NLP. Pretty exciting, eh? Why are we doing this, you might ask? Well, in addition to having fun, we want to expose you to another platform and demonstrate the benefits of experimenting with different AI systems. Kick the tires, so to speak. We already know that Alexa is an amazing intelligent voice platform, but API.AI is powerful in its own right and we want to introduce you to some of its core features.

We’ll be building another “Hello, World” project using API.AI as the NLU/NLP. In addition, we’ll build a Universal Windows Platform (UWP) app that will handle the voice interactions (record and playback) between you and API.AI. While our app will not have the power and sophistication that Alexa offers out of the box, we will be able to create robust, highly customized voice interactions using our UWP and API.AI. Lastly, the most important takeaway from this effort ...