Getting started with Kalliope
If you are using kalliope from a Rpi, the idea would be to configure your assistant from your main computer with an IDE and then push your config folder into your Rpi.
When you start kalliope using the CLI (
kalliope start), the program will try to load your
brain.yml in the following order:
- From your current folder, E.g
- From the default
brain.ymlwhich are located in the root of the Kalliope project tree.
This is a common tree of a Kalliope configuration folder:
kalliope_config/ ├── brains │ └── included_brain.yml ├── brain.yml ├── files │ └── kalliope-EN-13samples.pmdl └── settings.yml
We made starter kits that only needs to be cloned, placed into the Rpi and launched. You'll find the whole list of available start kits on the Kalliope's website. Those repositories provide you a structure to start playing and learning basics of Kalliope. Download the starter kit of your choice and open the folder with your IDE.
All files are expressed in YAML format (see YAML Syntax) and has a minimum of syntax, which intentionally tries to not be a programming language or script, but rather a model of a configuration or a process.
Let's open the main brain file of the English starter kit. You'll see there are some included sub brains file.
- includes: - brains/say.yml
- name: "say-hello-en" signals: - order: "Hello" neurons: - say: message: "Hello sir"
Let's break this down in sections so we can understand how the file is built and what each part means.
Items that begin with a
- are considered as list items. Items have the format of
key: value where value can be a simple string or a sequence of other items.
At the top level we have a "name" tag. This is the unique identifier of the synapse. It must be an unique word with the only accepted values : alphanumerics and dash. ([a - zA - Z0 - 9-])
- name: "Say-hello"
The first part, called signals is a list of input actions. You can add as many signal as you want in the "signals" section. If one of them is triggered, the neuron list will be executed.
signals: - order: "say-hello"
In the following example, we use just one signal, an "order", but it an can be:
- an order: Something that has been spoke out loud by the user.
- an event: A date or a frequency (E.G: repeat each morning at 8:30)
- a mqtt message A message received on a MQTT topic
- a geolocation From the position of your smartphone
- a community signal E.g: GPIO signal allow you to trigger actions from a button
- No signal. Then the synapse can be only called from another synapse or by the API
Then we have the neurons declaration. Neurons are modules that will be executed when the input action(signal) is triggered. You can define as many neurons as you want to the same input action (for example: say something, then do something etc...). This declaration contains a list (because it starts with a "-") of neurons
neurons: - neuron_1_name - neuron_2_name - another_neuron
The order of execution of neurons is defined by the order in which they are listed in neurons declaration.
Some neurons need parameters that can be passed as arguments following the syntax bellow:
neurons: - neuron_name: parameter1: "value1" parameter2: "value2"
Note here that parameters are indented with one tabulation bellow the neuron's name (YAML syntax requirement).
In this example, the neuron called "say" will make Kalliope speak out loud the sentence in parameter message.
Neurons can be Core (installed by default) or community based (need to be installed).
Time to start Kalliope. Move into the folder and then start Kalliope:
cd /path/to/the/starter_kit kalliope start
Note: Do not start Kalliope as root user or with sudo
Kalliope will load settings and brain, the output should looks the following
Starting event manager Events loaded Starting Kalliope Press Ctrl+C for stopping Starting REST API Listening port: 5000 Waiting for trigger detection
Then speak the hotwork out loud to wake up Kalliope (with the right pronunciation depending on your starter kit. "Kalliopé" in french, "Kalliopee" in English, etc..). If the trigger is successfully raised, you'll see "say something" into the console.
Then you can say "hello" and listen the Kalliope response.
Say something! Google Speech Recognition thinks you said hello Order matched in the brain. Running synapse "say-hello" Waiting for trigger detection
That's it! You are ready to customize your assistant!