One of the most important features necessary to develop speech recognition and synthesis in a given language is a comprehensive annotated speech corpus in the language.
At present there are two corpora being buit. One is made up of a number of Maltese texts based on newspapers, news items and books. This is available from Mike Rosner's home page at the department of Computer Science and AI.Maltese Text Corpora
The other speech corpora available is made up of a speech, sentence, and phoneme database.
At present the phoneme part is still under construction. Eventually it is hoped that the annotation includes where in the continuous spoken sentences, the phonemes start and finish. This is research that is taking place now in final year projects with my students. The corpus is taken from read news items from radio malta. I am grateful to them for making the tapes avilable to us.
Each item is defined as a sentence in words, the sentence in phonemes, and the speech as a .wav file suitable for standard sound cards on PC's.
Click here if you want to access the speech corpus.