Grant Amount: US$ 29,790

Keywords: TEXT-TO-SPEECH, VISUALLY IMPAIRED, MONGOLIA

Geographic coverage: Mongolia

Objective

The objective of this project is to develop a Mongolian text-to-speech (TTS) converter and a simple human computer interface that is suitable for visually impaired people.

Research context

TTS conversion tools offer visually impaired people the ability to access and use personal computers and software. Currently there is no TTS software for visually impaired people in Mongolia. This project aims to develop a TTS converter tool with a simple computer interface in local language that meets the needs of visually impaired people and provides a foundation for further development of TTS applications.

Target beneficiaries

This project benefits visually impaired people, general users of computer systems, students and researchers and Mongolian information technology (IT) companies. The Mongolian TTS synthesizer will be available online and released under a General Public License (GPL).

Outputs

• Mongolian TTS converter;
• Mongolian character recognition tool;
• TTS converter software package for visually impaired people; and
• A user manual in Mongolian.

Research results and outcomes

Developing the Mongolian TTS system requires expertise in information and communications technology (ICT) and research into Mongolian language and related theories. During research on the technical aspects of the project, the team identified key work on research and development of TTS by the MBROLA project (http://tcts.fpms.ac.be/synthesis/mbrola.html) and Festvox project (http://www.festvox.org). The MBROLA project concentrates on academic research for speech synthesis, particularly on prosody generation, known as one of the biggest challenges for developers of TTS synthesizers and the Festvox project provides technical tools for TTS. The work of these projects was cited as useful to the development of the Mongolian TTS system, although the project does not simply replicate them.

The project also undertook research to further its understanding of Mongolian language, which is a branch of Altaic and has different phonetics, structure and grammar than western languages. The project conducted relative studies on existing Mongolian language theories and held consultations and formal meetings with leading linguistic theorists in Mongolia, in particular with members of the Department of Stage Speech Art of the Mongolian University of Arts and Culture. The objective was to identify potential conflicts and discuss possible solutions regarding existing language theories and their practical use in TTS. Two main theories for pronouncing Mongolian words (developed by Sh. Lavsanvandan, D.Sc and M. Bazarragchaa D.Sc) are being used to obtain consistent and practical approach in developing Mongolian TTS converter. The grammar theory of Ts. Damdinsuren was also found to be useful. In this theory, words are split into syllables. To process syllables into TTS, research was performed on how to form words and program correct pronunciation. This involved defining rules for tones, unpronounced syllables and characters and over 700 frequently used abbreviations. In addition 400 commonly used foreign words were identified and are being integrated into the system.

At the time of writing, the project had completed the following activities. The necessary linguistic research activities for the development of Mongolian TTS converter were complete. The team had developed a software package for identifying the different syllables that occur in Mongolian lexis and storing them in a database. An additional software package was developed for collecting abbreviations. The alpha version of the Mongolian TTS converter engine has been developed and is able to read those words and syllables registered and recorded in the database. The sound and mouse click driven interface of the Mongolian TTS converter tool has been designed. The database is being populated with recorded sounds, syllables, abbreviations and foreign words.

The following activities have been conducted to promote the activities and results of the project. The project’s website (http://www.infocon.mn/tts) has been developed and is being regularly updated with the results from the Mongolian TTS converter tool project, its source code and other relevant outputs and information. Plans are underway to cooperate with OpenMN (http://www.openmn.org) and the Mongolian Unix User Group (http://www.mnbsd.org) for wider dissemination of project outputs after completion. The project will distribute the TTS converter package on compact disks and through free download from the project website. Once project team reaches its desired result, it is planning discussions with the Ministry of Education and Culture to investigate further cooperation in distributing the TTS system.

Duration

Start Date: September 2005
End Date: September 2006
Total Duration: 13 Months

Contact information

Batpurev Batchuluun
Executive Director
InfoCon Co., Ltd
PO Box 24, Ulaanbaatar 210646, Mongolia
Telephone: +976 9909 1465
Fax: +976 1132 8469
Email: batpurev@infocon.mn, info@infocon.mn
 
Website: http://www.infocon.mn/tts/

Reference websites

http://tcts.fpms.ac.be/synthesis/mbrola.html
http://www.festvox.org
http://www.openmn.org
http://www.mnbsd.org