View article

Recent advances in the processing capabilities of personal digital assistants (PDAs) have enabled the creation of an end-to-end speech translation system which is completely hosted on an off-the-shelf PDA. Specifically, the system presented includes an HMM-based large vocabulary continuous speech recognizer using statistical n-grams, a translation system, and, a speech synthesis system that outputs speech in the target language. This paper describes the detailed architecture of our system and the functionality of its main components, focusing on the optimization efforts employed to achieve real time results on such a limited platform. In a preliminary evaluation, the PDA system is found to achieve comparable performance with the corresponding system developed for desktop computers.