Greek Forced Alignment: Assessing the Accuracy of the Montreal Forced Aligner

Anonymous

Abstract: Forced alignment has allowed for the rapid creation and annotation of corpora. In this study we examine the Montreal Foreced Aligner and its accuracy of aligning Greek data. Using a conversational Greek corpus we train a small grapheme-to-phoneme model and use this model to align the entire corpus. We compare our results to various previous studies of the MFA and other forced alignment software and conclude that forced alignment greatly increases the ability to create new corpora for low-resource and understudied languages.