Aid-SLT is a multilingual spoken language translation system tailored for humanitarian settings, designed to help in situations where primary data collectors and informants do not share a common language.
The idea of using automatic speech translation in humanitarian settings stems directly from the work done by the University of Geneva on MedSLT, an analogous tool intended for use in doctor-patient diagnosis dialogues. MedSLT has been developed in a hand-held mobile version and supports English, French, Japanese, Spanish, Catalan and Arabic.
Communication using Aid-SLT would be as follows:
The data collector presses on the recognition button and speaks into the microphone; he can then check the back-translation of his utterance (to verify that what he said was well understood by the system). If the data collector accepts the produced string, it gets synthesized to the informant. The informant presses on the recognition button to speak a direct answer to the question; alternatively, to produce an answer he can make use of the help window that displays a set of potential answers.
For more information:
http://www.issco.unige.ch/en/research/projects/medslt