brock / voicebox Goto Github PK

Download

Press alt+s to trigger the voicebox window

VoiceBox currently has a very limited set of sentences it understands. If you want to contribute to this, please look at the adding responders section.

Hello
What time is it?
My name is Thomas
What's my name?
Tell me about [any topic]
What is a guitar
What's the weather like in Dublin

You can change the voice in the OS X speech and dictation settings. In the system voice dropdown you can select 'custom' which will allow you to download more voices. I use 'Samantha' which is the same voice used by siri on the iPhone.

VoiceBox is built on top of electron.

npm install
electron .

Now press alt+s

Create a new javascript file for your responder in the /responders directory. You can then use the voiceBox.addResponder method to define your responder.

The addResponder method takes two parameters:

• regex ( Regex ): The users input will be matched against this to determine if it should respond
• response ( Function ) : If the regex matches the input the response function will be called

The voiceBox.respond(message) method is then used inside the response function to output a response.

NB: The response function is only called if it is the last responder to match the input

voiceBox.addResponder(/Hi|Hello|Hey/i, function(){
  voiceBox.respond("Hello")
})

voiceBox.request can be used to make external API requests.

NB: This uses the request module

voiceBox.addResponder(/What's the weather like in (.*)/i, function(){
  voiceBox.request('http://api.openweathermap.org/data/2.5/weather?q='+RegExp.$1, function (error, response, body) {
    if (!error && response.statusCode == 200) {
      voiceBox.respond(JSON.parse(body).weather[0].description)
    }
  })
})

In some cases a responder may require the user to give more information. This can be done using the voiceBox.addChoice and voiceBox.choice methods.

First, you need to add a choice using the voiceBox.addChoice method. This works the exact same way as the voiceBox.addResponder method.

You can then respond asking for more information using the voiceBox.choice(message) method. This tells VoiceBox that we are asking for more information and will begin listening again after output of the message has finished.

voiceBox.addResponder(/give me options/i, function(){
  voiceBox.addChoice(/one|1/i, function(){
    voiceBox.respond('You selected option one')
  })

  voiceBox.addChoice(/two|2/i, function(){
    voiceBox.respond('You selected option two')
  })

  voiceBox.choice('Do you want option one or option two?')
})

You can store information using the following methods

voiceBox.db.set('users_name', 'thomas')
voiceBox.db.get('users_name')

All the things

• Better application structure
• Improve documentation
• Ability to change config to pull in responses as npm packages ( similar to how Hubot handles scripts )

## License VoiceBox is released under the MIT License.

The duckduckgo responder uses their awesome API