An automatic and adaptive phonetic transcriber for the Basque language

摘要

This paper proposes an automatic and adaptive phonetic transcriber (PT) for Basque. In order to develop this transcriber, the sociocultural situation of the Basque Country has to be taken into account. Basque is an minority language that has many dialects, so it shows a wide range of language varieties spread across very small geographical zones. Thus, words can be pronounced in different ways depending of the origin of the speaker, or if she/he is non-native, which is very common due to the sociolinguistic characteristics of the Basque region. In this context, there are phonetic realizations out of transcription standards for Basque, so we present an automatic PT that is able to achieve transcriptions in standard Basque and adapt to real particular realizations. The phonetic transcription of spoken discourse is the foundation on which any system based on voice recognition is built, so this point is crucial not only to enrich the knowledge of a minority language, but also to develop other related speech recognition applications.