This exploratory research seeks to demonstrate that new approaches to Automatic Speech Recognition (ASR) and Machine Translation (MT) permit automatic, real-time translation well enough for emergency triage. In any major disaster, such as a hurricane, there will be a huge number of 9-1-1 emergency calls in a short period of time; many will be in languages other than English. Even in the best of times, there is a chronic shortage of translation for triage of non-English calls at many U.S. 9-1-1 centers.
For emergency triage, full recognition and translation are unnecessary; only the type of emergency and relevant details, such as location, need be recognized. Nevertheless, new advanced ASR methods are needed to handle the difficult characteristics of non-English 9-1-1 calls, which contain emotionally-distressed speech in noisy environments and may include English phrases mixed in with the other language. For ASR, the exploratory system uses multilingual acoustic models enhanced by articulatory features, and multilingual grammars combined with n-gram language models, to recognize speech. As for MT, the exploratory system only classifies the ASR outputs into ``Domain Actions'', categories relevant to triage such as ``Request-Ambulance'' or ``Report-Flooding''. The exploratory system works with Spanish, but the approaches used are applicable to any language, and transferable to other task domains.
Demonstrating the feasibility of speech translation technology for serving 9-1-1 call centers should enable follow-on projects to better serve all non-English speakers during major disasters. We expect to make the resulting transcribed and annotated speech corpus available for other researchers.
