Interactive ChatBot in Turkish

Query: sen güzel bir kızsın.
ChatBot: sen daha güzelsin!!

we trained seq2seq model on Turkish twitter data to build a chatbot, and we built an interface(human head) to intract with the chatbot.

Prerequisites

• python3
• TensorFlow <= 1.11.0
• TensorLayer >= 1.6.3
• in addition you have to be connected to internet

How it works

This repository contains two main parts, trigger word part (triggerWord folder) and chatbot part (chatbot folder)

• Trigger Word Detection

  Uyan Ay ===>

  

  Here is come the first part the trigger word part! In order to interacte with human interface and speak with it! you have to wake it up.

  we trained our bot to wake up whenever it hears 'uyan ay', it is similar to amazon Alex when it hears 'Alex' trigger word, it wakes up.

• ChatBot

  our bot is task-specfic chatbot it has one task called general conversation, what it does is for any query text given to it (ex: question, jokes, ..) it repoenses with other text(answer,make statement), derived and similir to twitter responses.

  first your voice will be converted to text using voice recognition api from google, then feeded to the text-to-text AI system finally the result will be printed to the human interface(as voice and text).

  query: seni çok seviyorum.

  ChatBot: ben de seni çok seviyorum yanımızda olduğun için sonsuz teşekkürler.

  

Instalation

• ubuntu

  if you are on ubuntu you should run first:

   $ sudo apt install libasound-dev portaudio19-dev libportaudio2 libportaudiocpp0 ffmpeg libav-tools
  
   $ sudo apt-get install python3-tk 

  after that run this:

    - Ubuntu python3:
    	pip3 install -r requirements.txt
  
    - Ubuntu python:
    	pip install -r requirements.txt
  

• windows

  I didn't test it on windows but it should be working fine.

    run: pip install -r requirements.txt
  

run the chatbot

Ubuntu:
	for python 3: 
		python3 my_robot.py
	for python 2:
		python my_robot.py

Windows:
	python my_robot.py