Roby is an open source AI chatbot, with Local speech recognition and speech synthesis system. released under GPLv3, more details see LICENSE.
Roby-0.0 is just a demo program fragment for testing some basic functions. It only works on Linux. Supports both English and Chinese. You can watch the demo at: http://v.youku.com/v_show/id_XODExMTU1ODQ4.html
How to use:
Before runnig Roby, make sure you have installed:
- python-2.7
- pyaiml-0.8.6---------------a python AIML interpreter
- pyaudio-0.2.8--------------python bindings for portaudio
- pyserial-2.7---------------this python module encapsulates the access for the serial port
- sphinxbase-5prealpha-------the basic libraries for CMU Sphinx speech recognition engine
- pocketsphinx-5prealpha-----the CMU Sphinx speech recognition engine
- festival-2.1---------------for English TTS output
- ekho-6.0-------------------for Chinese TTS output
Python modules can be used pip or easy_install to install. Ubuntu users can use "apt-get" to install festival and ekho. sphinxbase and pocketsphinx can be compiled from source code. You may need the web pages below: http://www.speech.cs.cmu.edu/sphinx/doc/Sphinx.html https://github.com/andelf/PyAIML http://people.csail.mit.edu/hubert/pyaudio/ http://people.csail.mit.edu/hubert/pyaudio/ http://www.cstr.ed.ac.uk/projects/festival/ http://www.eguidedog.net/ekho.php
- Before running Roby, you must change the value "ttyPath" in start.py main function.
- The first line in start.py "# encoding: uft-8" can not be deleted, or it will get problems with Chinese functions.
- Roby must work with "sudo", or tty port can not be used.
- Use "sudo python start.py" to start Roby.
- You should firstly say: "Roby", then Roby says: "yes, sir", then you can say the command words.
- If Roby's respond end with "?", he will continue listen your command words without saying "Roby".
- The 0.0 version only supports several commands below:
- how are you?
- fine
- what's the time?
- what's your name?
- how old are you?
- chinese mode
- write serial port
- exit
- 几点了？
- 英文模式
You can build your own test speech recognition models under the folder names "model".
Any questions, email to ideamark@qq.com