Modélisation acoustico-phonétique de langues peu dotées : Études phonétiques et travaux de reconnaissance automatique en luxembourgois
Martine Adda‐Decker, Lori F Lamel, Gilles Adda · HAL (Le Centre pour la Communication Scientifique Directe) · 2014
Luxembourgish, a Germanic-Franconian language, is embedded in a multilingual context on the divide between Romance and Germanic cultures and remains one of Europes under-described languages. This paper investigates the similarity between Luxembourgish phone segments with German, French and English via forced speech alignment techniques. Making use of monolingual acoustic seed models from these three languages, as well as multilingual models trained on pooled speech data we investigated whether Luxembourgish was globally better represented by one of the individual languages or by the multilingual model. While globally, the German models provide the best match, a phone-based analysis, shows language-specific preferences. First ASR results illustrate the accuracy of the various sets of monolingual and multilingual acoustic models and Luxemburgish acoustic models built from 1200 hours of untranscribed Luxemburgish audio data using unsupervised methods.