[0:01] Marvin, turn on the lights
[0:07] “OK”
[0:08] Marvin, turn off the bedroom
[0:14] “OK”
[0:16] In this tutorial, we build our very own Alexa using an esp32

[0:21] and the Wit.ai service from Facebook
[0:25] We’re going to cover two main topics:
[0:27] Wake Word Detection
[0:29] and Intent Recognition
[0:31] Finally, we’ll wire these two together to create our Alexa system.
[0:36] For our wake word detection, we’ll be training up a model using TensorFlow

[0:40] and then running that model using TensorFlow Lite on our device.
[0:44] We’ll cover: collecting and preparing our training data, training the model,

[0:50] and converting the model to TensorFlow lite.
[0:53] For recognizing the user’s intent we’ll be using Wit.ai from Facebook.
[0:58] We’ll create a new Wit.ai application, create the intents entities and traits for our Alexa system,

[1:06] and then train our application to recognize a set of example phrases.
[1:11] Finally, we’ll wire all of this up into our Alexa system and run it on an embedded device.
[1:16] We’ll cover: audio input and output, running the neural network,

[1:21] streaming audio to Wit.ai and processing the results from Wit.ai.
[1:27] With these steps, we’ll create a complete working DIY Alexa
[1:33] Marvin, tell me a joke
[1:39] “What goes up and down but does not move?”
[1:43] “stairs”