Home

Awesome

Tell the Time

Build a robot that can tell the time in different cities with Watson

"Watson, what time is it in Berlin?"

This module provides Node.js code to get your Raspberry Pi to tell the time in any city you choose to tell Watson about. It uses Watson Speech to Text to parse audio from the microphone, uses Watson Conversation to generate a response, and uses Watson Text to Speech to "read" out this response!

This will only run on the Raspberry Pi.

How It Works

Hardware

Check out [this instructable] (http://www.instructables.com/id/Build-a-Talking-Robot-With-Watson-and-Raspberry-Pi/) to prepare your system. You will need a Raspberry Pi 3, Microphone, Speaker, and the TJBot cardboard.

Build

We recommend starting with our [step by step instructions] (http://www.instructables.com/id/Build-a-Talking-Robot-With-Watson-and-Raspberry-Pi/) to build this recipe.

Get the sample code and go to the application folder. Please see this instruction on how to clone a repository.

Install ALSA tools (required for recording audio on Raspberry Pi)

sudo apt-get install alsa-base alsa-utils

Install Dependencies

npm install

Set the audio output to your audio jack. For more audio channels, check the config guide. 

amixer cset numid=3 1    
// This sets the audio output to option 1 which is your Pi's Audio Jack. Option 0 = Auto, Option 2 = HDMI. An alternative is to type sudo raspi-config and change the audio to 3.5mm audio jack.

Update the Config file with your Bluemix credentials for all three Watson services.

edit config.js
enter your watson usernames, passwords and versions.

Creating a Conversation Flow

You need to train your robot with what to say and when to say it. For that, we use [Watson Conversation] (https://www.ibm.com/watson/developercloud/conversation.html). Open a browser and go to IBM Watson Conversation link From the top right corner, select the name of your conversation service and click 'create' to create a new workspace for your robot. You can create intents and dialogs there. Here is a step-by-step instructions to create a conversation flow.

Creating intents

An intent for telling the time might look like this:

tellTheTime intent

Creating dialogs

A Dialog node for telling the time for different timezones might look like this:

tellTheTime dialog

The response should contain a placeholder (todays_date) that the node application can replace with a javascript Date. When a specific city entity is found, the response for each city should return the timezone offset in the output context:

timezone in offset output

Running

Start the application

node app.js

Then you should be able to speak to the microphone. The robot gets better with training. You can go to your Watson conversation module to train the robot with more intents and responses.

Monitoring

The app serves up a Monitoring web interface locally on port 3000. You should be able to point to this from another device on the same network as your Rasberry Pi. The Monitoring UI allows you to stop and resume listening on TJBot's microphone. You can also tell whether the TJBot module has stopped due to an unexpected error.

tellTheTime monitor app

The logs show the REST calls made from the Monitoring UI to control TJBot:

TJBot is listening, you may speak now.
TJBot monitor app listening on port 3000!
Fri, 24 Mar 2017 22:38:53 GMT LOG Router request GET /tjbot
Fri, 24 Mar 2017 22:38:54 GMT LOG Router request GET /tjbot/status
Fri, 24 Mar 2017 22:39:00 GMT LOG Router request POST /tjbot/pause
Fri, 24 Mar 2017 22:39:19 GMT LOG Router request POST /tjbot/resume
TJBot is listening, you may speak now.

Customization

The attention word is the word you say to get the attention of the robot. The default attention word is set to 'Watson' but you can change it from config.js. Some words are easier for the robot to recognize. If decided to change the attention word, experiment with multiple words and pick the one that is easier for the robot to recognize.

The default voice of TJBot is set to a male voice (en-US_MichaelVoice) but you can change it from config.js. Two female voices are available for TJBot (en-US_AllisonVoice and en-US_LisaVoice).

// The attention word to wake up the robot.
exports.attentionWord ='watson';

// You can change the voice of the robot to your favorite voice.
exports.voice = 'en-US_MichaelVoice';
// Some of the available options are:
// en-US_AllisonVoice
// en-US_LisaVoice
// en-US_MichaelVoice (the default)

Dependencies List

Contributing

See CONTRIBUTING.md.