As part of our Custom Development efforts, we always aim to take on new challenging projects. Crimsonwing’s most recent project driving innovation is the custom development of a speech synthesiser for the Maltese language – a first of its kind. The speech synthesiser will enable computers to produce and process spoken Maltese, which will allow a more natural interaction between the speaker and the computer, recognising keyboard and verbal input and being able to transform text into audio speech format. The first prototype was delivered in August 2010.

This project was initiated by the Foundation for Information Technology Accessibility (FITA), and is co-financed via an 85% grant by the European Regional Development Fund and 15% by the Maltese government. Being EU funded, the project falls under the EU’s cohesion policy 2007-2013 “Investing In Competitiveness For A Better Quality Of Life”. Crimsonwing won the tender for this project in February 2010 and Carmel Gafa, Head of Technology, said “We are proud to have won the tender for this challenging and interesting project. The synthesiser will make a significant difference to many Maltese speaking people who need accessible electronic media, for example partially sighted, blind, illiterate, injured and physically disabled people. It is also an honour to work with leading experts in this field, like Prof. Paul Micallef and Prof. Ray Fabri, and we are making excellent progress on the project.” Based on a strong commitment of social responsibility, Crimsonwing is very dedicated to this project with a firm ambition to continuously improve social environments.

The technology for speech recognition engines already exists but not for the Maltese language. Therefore, people requiring speech-enabled technology have had to rely on alternative languages. One of the challenges with this project is the Maltese language itself, as pronunciation and grammar rules are complex. Therefore, development within the highly specialised field of speech synthesis is a very challenging and rewarding endeavour for an IT company.

A key requirement for the synthesiser, which is developed for the Microsoft Windows platform, is compliance with Microsoft’s Speech Application Programming Interface (SAPI). This is a software standard that ensures compatibility with a wide variety of assistive and educational software applications available on the Windows platform. The software supports Windows XP, Windows Vista and Windows 7 32bit and 64bit editions.The Maltese Speech Synthesiser is being developed as a series of prototypes culminating in a finalised product that is slated for release in July 2012. It will be available for free download from a specifically developed website, or alternatively a CD with the program can be requested from FITA.

The first prototype, presented in August 2010, was a rudimentary system featuring a male voice, and based on academic research carried out at the University of Malta. This prototype was not intended for further development, but served Crimsonwing on two fronts: to build a clear picture of the development requirements, and to cement confidence in Crimsonwing’s abilities to deliver the final software.

The second prototype, presented in March 2011, was architected from scratch to serve as the basis for the final product. It featured improved language analysis for proper handling of ambiguities in the text, and higher quality audio output, based on a new male voice recorded in a professional sound studio. The language analysis process is driven by a lexicon: a comprehensive word dictionary that contains information such as the exact (phonetic) spelling of each word, the word’s part of speech, syllabication and stress information. Each voice actor was required to read a 15,000 word script, specially compiled from an initial 20 million word corpus, statistically representative of the sounds in the Maltese language. During development of this prototype, a sound processing tool was specially developed to enable Crimsonwing to build a sound database for each voice from the recorded script renditions.

The third prototype, presented in October 2011, featured a larger lexicon and sound database for the male voice, and the introduction of a female and child voice databases. Jeffrey Bezzina, Project Manager at Crimsonwing commented “It’s thrilling – hearing the male, female and child voices, coming out of the new Maltese Speech Synthesiser tool. Testers are already choosing their preferred voice.”

This prototype also saw the introduction of an intonation model of the Maltese language, resulting in more natural sounding speech, and the verbalisation of numbers, emails and other such types of textual representations. As explained by Roger Davies-Barrett, Project Manager for ERDF114 “the aim has always been to get as close to natural speech as possible and you only have to listen to the English engines trying to pronounce Maltese text to realise how much better the Synthesiser we’ve developed is. Many e-government websites will now be fully accessible to Maltese speaking people who previously gave up due to badly pronounced Maltese.”

The final system will eventually feature fully compiled sound databases for the three voices, an even larger lexicon, improved language processing and control of volume, speaking rate and pitch. The software will also undergo a series of performance tuning exercises to improve the overall efficiency and responsiveness of the synthesiser.

Maltese Speech Synthesiser - Detailed (154.21 kB)

 FITA Document - Speech Synthesiser (404.92 kB)

Here are some previous News Items that we published about the Speech Synthesiser:

News Item 1- Maltese Speech Synthesiser

News Item 2 - Good Progress on Project

Here is our latest Press Release on the topic - brand new!

Maltese Speech Synthesiser raring to go - Launch event and Seminar coming up this summer

Please see the Custom Development section for other projects.