This paper presents the methods used and results obtained for the creation of a Festival-compatible pronunciation dictionary of above 10k words for the kabyle language. Kabyle is a berber dialect spoken in Northern Algeria. This dictionary will be useful in the design of text-to-speech and automatic speech recognition systems for the kabyle language. It was built using a bootstrapping method in which we incrementally build rules to predict word pronunciations while correcting wrong predictions. We thus obtain a large pronunciation dictionary as well as a set of rules to predict pronunciations for unknown words. The rules are embedded in Classification and Regression Trees and achieve 91,62% of correct prediction rate for entire words and 97,85% for phonemes.
CITATION STYLE
Lyes, D., Leila, F., & Hocine, T. (2019). Building a pronunciation dictionary for the Kabyle language. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11658 LNAI, pp. 309–316). Springer Verlag. https://doi.org/10.1007/978-3-030-26061-3_32
Mendeley helps you to discover research relevant for your work.