2020/04/21 by Aryaman Arora, Arora, Aryaman, Luke Gessler +3
Computer Science · #Computation and Language (cs.CL) #FOS: Computer and information sciences #I.2.7 #Music and Audio Processing #Natural Language Processing Techniques #Speech Recognition and Synthesis
paper · pdf · doi:10.48550/arxiv.2004.10353
openalex publication_date 2020/04/21 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
Hindi grapheme-to-phoneme (G2P) conversion is mostly trivial, with one\nexception: whether a schwa represented in the orthography is pronounced or\nunpronounced (deleted). Previous work has attempted to predict schwa deletion\nin a rule-based fashion using prosodic or phonetic analysis. We present the\nfirst statistical schwa deletion classifier for Hindi, which relies solely on\nthe orthography as the input and outperforms previous approaches. We trained\nour model on a newly-compiled pronunciation lexicon extracted from various\nonline dictionaries. Our best Hindi model achieves state of the art\nperformance, and also achieves good performance on a closely related language,\nPunjabi, without modification.\n