University of West Bohemia is the only one university in the Czech republic that offers its applicants a service that allows to acquire pieces of information about the results of the entrance exams. There are two possibilities how to obtain results of entrance exams. Either visiting the web pages of the university or calling to the information system. This servise was available nonstop and there were mobile numbers affordable too. After calling the one of the phone numbers, the telephone machine asks for identification first of all. After registration into the system user has to choose the faculty and the specialization which he/she wants to know the results of the entrance exams about. Then the user listen to the results affordable in the time of calling.
The developed system is able to recognize and process user voice input and communicate using synthesized voice. When the user calls for the results, he/she is greeted with a short introduction of system usage. The full help is available in any time after touching a help key or saying help. After introduction, the system asks user to enter his/her identification number (ID). The number can be entered using a dial pad or verbally. If the ID is valid, the user is prompted to choose the faculty (provided that he/she took the entrance examinations for more than one faculty) and the field of study. The list of fields of study could be quite long and forcing the user to listen to the entire list without having a chance to react could results into users boredom and dissatisfaction. Thus we have decided to use the so-called point and speak method. The essence of this method is to switch the speech engine to the recognition mode for approximately one second after each item of the list. The user is instructed to say a short confirmation phrase if he/she wishes to select the item that was just presented. The content of the confirmation phrase is not important since the system only detects a speech activity and does not try to decode the words itself because of the problems connected with the inevitable delay occuring during the switching between synthesis and recognition mode. The user is of course shielded from these technical details and is encouraged to use the confirmation phrases like Yes and That one. After collecting all the necessary input, the system presents the entrance examination results found in the university database.
Technical solution requires a computer that is hard-wired to the net and to the telephone net at the same time. The computer is connected to the telephone net by special card which allows processing of more phone calls at the one time. Basis of the system is VoiceXML interpreter. The application written in this language is easily modifiable and extensible. VoiceXML document (can be generated by server side scripting language) is processed by interpreter and his content is presented to the user. User assigns information (handle the system) by DTMF. VoiceXML interpreter is the central program that has to communicate with the user (with assistance of a module for communication with phone lines), listen to the user what it says or assign and it has to convert arbitrary text to speech. The system was developed with help of Centrum of applied cybernetics, Department of Cybernetics at Faculty of applied sciences in the University of West Bohemia and the SpeechTech company.
From the year 2004 the system serves 12 lines simultaneously.
|2003||3 497||2 418||74:26|
|2004||3 211||2 336||78:22|
|2005||2 917||2 063||85:15|
|2006||1 980||1 287||82:18|
At this place I have to thank for support to: Frantiąek Král (phone lines), Tomáą Kotouč, Petr Grolmus, Jakub Urbanec (document server), Jan Rychlík (HW equipment - PC Dual XEON, ISDN Controller C4), and Department of Cybernetics and SpeechTech company (Intepreter VoiceXML).
Detailed information, and comments: Luboą ©mídl