A company originally established in the UK is developing the first Maltese speech synthesiser, which will produce spoken Maltese by recognising keyboard and verbal input and transforming text into audio speech.
The work is being carried out entirely in Malta by a consortium of Maltese entities. The release of the synthesiser is scheduled for the end of March 2012.
The company is Crimsonwing, whose Text To Speech system will include a lexicon of the Maltese language compiled from a 20 million-word corpus. The current lexicon stands at around 15,000 words.
Although similar software tools have been in existence for some time, a speech engine for the Maltese language is not yet available.
One of this project’s main challenges is the complex pronunciation and grammar rules embraced by the Maltese language.
This software, which will have three different voices – male, female and child – could facilitate access to Maltese electronic text for readers with physical, sensory or cognitive requirements. It will also offer speech control features like tone, pitch and rate of speech.
Crimsonwing senior technical architect Colin Vella said members of the Maltese academia in the field of linguistics and sound engineering have also shown a lot of interest in the project and it may therefore serve as a foundation for further research.
This EU and government funded project was initiated by the Foundation for Information Technology Accessibility (FITA), with the intention of enabling visually impaired people and those with difficulties including dyslexia and illiteracy, with a facility for reading Maltese digital documents through the use of speech-enabled software capable of processing Maltese text.
Mr Vella said the development team was treating the speech synthesiser as an accessibility and educational tool.
“However, the software may find other applications, such as educational games and speech therapy. Indeed, the software has also been earmarked for use within the education sector, which is also being involved in the software’s test cycles,” he said.
Crimsonwing won the project tender in February 2010. The company, established in 1996 in the UK as Magus International, set up a fully owned Malta centre in 1997.
The synthesiser conforms to Microsoft’s Speech Application Programming Interface (SAPI), enabling the software to be installed as an additional speech service on Microsoft Windows XP, Vista and Windows 7 so that any speech-capable software could make use of these new Maltese voices.
Mr Vella said the pioneer in this field was Dr Ing Paul Micallef from the Department of Communications and Computer Engineering at the University of Malta.
As part of his seminal 1997 PhD thesis, A Text To Speech System, Prof. Micallef implemented a prototype synthesiser that was capable of intelligible utterances but had mechanical sounding speech.
Other researchers, Mr Vella said, built upon this work.
The consortium is made up of the research and development team at Crimsonwing which is acting as the software development team, Prof. Micallef as the primary consultant, Prof. Ray Fabri from the Institute of Linguistics, as the leading linguistics consultant, and Prof. Ing Kenneth Camilleri from the Department of Systems and Control Engineering as the leading consultant on digital signal processing algorithms.