Getting started

This will help you get up and running with Jenova.

requirements

The following packages need to be installed via pip

• gunicorn: lightweight WSGI HTTP Server
• falcon: super fast RESTful framework
• httplib2: HTTP client

For the bot, these packages are required:

• pygame: for audio playback
• pyttsx, gTTS: TTS engines

For the server, these packages are required:

• sklearn, nltk: for machine learning algorithms and utilities
• numpy, scipy: libraries used by sklearn
• tinysegmenter: Japanese tokenizer

Python3 and ev3dev-lang-python are also required to run the application. It's ok to run the application without ev3dev, but you won't have the features related to Lego EV3 like motor controlling.

installation

1. Install the dependencies

It depends on the operating system you are using, here are the example for Ubuntu 17.04

sudo apt-get install python3 python3-pip python3-ev3dev
sudo pip3 install gunicorn falcon httplib2 pygame pyttsx gTTS numpy scipy sklearn nltk tinysegmenter

If you want to use mary-tts as TTS engine, it must be installed separately: https://github.com/marytts/marytts/wiki/Local-MaryTTS-Server-Installation

(Note that there are some problem with installing and running pyttsx with Python3. Checkout this fork of pyttsx instead)

2. Checkout the source code and install the framework

git clone --recursive https://github.com/dungba88/jenova.git
cd jenova
sudo pip3 install orion/

3. Run the bot

gunicorn main --chdir bot/ -b 0.0.0.0:8081 --reload
gunicorn main --chdir server/ -b 0.0.0.0:8080 --reload

(You may need to run with nohup command)

Now the bot can be accessed from http://localhost:8081 and the server can be accessed from http://localhost:8080

4. Start the UI

For this, you need to install a web server which supports static files, like Apache or nginx, and make the ui/ folder accessible by HTTP. Setup will depend on which web server you choose.

Now you are ready to interact with bot!