The MBRDICO Project
What is MBRDICO?
MBRDICO is a talking dictionnary using MBROLA as a back-end speech synthesizer. Text processing is performed using a complete GNU GPL package for automatic phonetization training (letter/phoneme alignement, decision tree building, stress assignment) and duration/intonation generation. At the moment American, Arabic, British English, Dutch, French and Spanish are available. This work is the result of a collaboration between:
Contributors are welcome to enlarge this set. If you are interested, email us.
- Faculté Polytechnique de Mons
- Carnegie Mellon University
- University of Edinburgh
If you want to understand how the MBRDICO letter to phoneme alogrithm works, read "LETTER TO SOUND RULES FOR ACCENTED LEXICON COMPRESSION", by Vincent Pagel, Kevin Lenzo and Alan W. Black Proc ICSLP98 Sydney.
What can I download?
Freely available dictionaries on the net to train a phonetizer:
The MBRDICO software (release 020708) is provided as a GNU GPL package with sources and precompiled binary for Windows95/NT and PC/Linux.
- The automatic learning ID3 software (release 020607) makes it possible for third parties to develop their own tree-based phonetizer, which can then directly be used with the MBRDICO software.
Many thanks to Kevin Lenzo (CMU) for his amazing Perl implementation of the ID3 algorithm that initiated this project.
And to Alan Black (CSTR)) for his help in testing different letter to phoneme + stress strategies.
I warmly thank :
- Nawfal Tounsi (FPMS) for providing an Arabic system
- Fabrice Malfrere (FPMS) for providing a Spanish and Dutch system
- Richard Beaufort (DEC2 student) who heavily corrected our French phonetic dictionary (his linguistics knowledge is welcome).
- Alain Ruelle and Michel Bagein (FPMS) for helping with C++/STL and the wicked Visual C++.
- Philippe Devallois for spotting bugs !
V. Pagel( mailto:firstname.lastname@example.org )
Last updated March 20, 2000, send comments to email@example.com