VERFAHREN ZUR MEHRSPRACHIGEN SPRACHERKENNUNG

摘要

A cross-lingual method for recognizing a sequence (30) of speech elements spoken by a first language native speaker. The method comprises the steps of: building native speech element models (36) for speech elements belonging to said first language, building cross-lingual speech elements (37) for speech elements belonging to a second language The cross-lingual speech elements (37) are built from a set of first language speakers. The native models (36) and the cross-lingual models (37) are both used for the recognition of the sequence.