Low-Resource G2P and P2G Conversion with Synthetic Training Data

摘要

This paper presents the University of Alberta systems and results in the SIGMOR-PHON 2020 Task 1: Multilingual Grapheme-to-Phoneme Conversion. Following previous S1GMORPHON shared tasks, we define a low-resource setting with 100 training instances. We experiment with three transduction approaches in both standard and low-resource settings, as well as on the related task of phoneme-to-grapheme conversion. We propose a method for synthesizing training data using a combination of diverse models.