A machine is gifted with a "talking" system if it is able to:
- receive word expressions from a human agent
- translate the received words and to compute a possible answer
- give the computed answer to the human agent in a way he can understand
These skills involves the following 3 sub-systems:
- Hearing system
- Knowledge database
- Speaking system
The Hearing System is an input system standing between the human agent and the Knowledge database. Usually, the input is represented by a text buffer which is sent to the Knowledge database. The text buffer might be provided or calculated. In a
text-based talking system, the text buffer is provided by the human agent, and it is sent to the Knowledge database unmodified. In a
voice talking system, the text buffer is calculated from the input voice, by using a
voice recognition system.
The knowledge database is an intermediary system: it receives a text buffer from the Hearing system as input and it provides a text buffer to the Talking system as an output. The output buffer is the computed answer; the computation might be implemented in a very simple way or a very complex one. A complex system involves a
memory system and an artificial intelligence system. The simple systems only contains IF statements, for example:
IF HumanAgentQuery ISLIKE "how are you"
MyAnswer IS "I'm doing well"The Speaking system gets the text buffer (answer) from the Knowledge database and provides it unmodified to the human agent (in text-based talking systems) or converts the text into a speech sound wave (in voice-based talking systems) by using a
Text-To-Speech System.
An example of voice-talking robot is presented here:
http://robotics.j2me.info/AIrobot.html Be sure you meet the minimum requirements in order to be able to run the applet program:
- Standard audio card, with attached speakers and microphone
- 512 MB of System RAM
- Web Browser like Internet Explorer or Mozilla familly
- Java Runtime Environement (see the tutorial on the page)
Enjoy!
