METHOD AND APPARATUS FOR THE PREDICTION OF MULTIPLE NAME PRONUNCIATIONS FOR USE IN SPEECH RECOGNITION

摘要

Multiple, yet plausible, pronunciations of a proper name are generated basedonone or more potential language origins of the name, and based further on thecontext inwhich the name is being spoken -- namely, on characteristics of the populationofpotential speakers. Conventional techniques may be employed to identify likelycandidates for the language origin of the name, and the characteristics of thespeakerpopulation on which the generation of the pronunciations is further based maycomprise,for example, the national origin of the speakers, the purpose of the speech,thegeographical location of the speakers, or the general level of sophisticationof the speakerpopulation. Specifically, a method and apparatus is provided for generating aplurality ofplausible pronunciations for a proper name, the method or apparatus for use inperforming speech recognition of speech utterances comprising the proper namebyindividuals within a given population of speakers, the method or apparatuscomprisingsteps or means respectively for (a) identifying one or more of a plurality oflanguages as apotential origin of the proper name; and (b) generating a plurality ofplausiblepronunciations for the given proper name, one or more of the plurality ofpronunciationsbased on the one or more identified languages, and the plurality of plausiblepronunciations based further on one or more characteristics associated withthe givenpopulation of speakers.