Speech_AI
Simple speech linguistic AI with Python
It supports almost any natural language. By default it works in russian language. if you want to change it, please check documentation of libraries below. It can be done easily in 3 fixes:
- Change comments
- Change language of recognizer and synthesizer and train bot with your language sources (corpus, Twitter, etc.).
Main script is speech_ai. It recognizes your speech, looking for answer by chatterbot library and replies you with speech synthesized by GTTS library. Because of Libraries specials, it needs internet connection.
This script uses:
You can use this idea in you video game or maybe with Smart House.
Enjoy!
Installation
Debian-based linux
At first you need to install libraries, listed above. For this I recommend use of python environments (like conda environments)
conda create --name speech_ai
source activate speech_ai
conda install python=3.5
# Install pyaudio
# You can try pip3 command with sudo if errors appears
sudo apt-get install python-pyaudio python3-pyaudio
pip3 install pyaudio
pip3 install gTTS
pip3 install SpeechRecognition
pip3 install chatterbot
pip3 install pygame
Windows XP and older
In Windows OS we have several methods to install packages:
- We can install compiler suggested for our Python version
- Or we can easily use
Wheel.
On windows it's bit difficult to installpyaudioandpygame. So, easy way to useWheel. This package versions for Python 3.4.x, but you can download versions what you need
- Install Python 3.4.x standalone or in Anaconda
- Download wheels depending on your architecture (x86 or amd64) and Python version:
- Install packages (install wheel if you haven't did it yet).
pip install --upgrade pip
pip install wheel
pip install pygame-1.9.3-cp34-cp34m-win32.whl
pip install PyAudio-0.2.11-cp34-cp34m-win32.whl
pip3 install gTTS
pip3 install SpeechRecognition
pip3 install chatterbot
Run
python3 speech_ai.py
