Better Morphology Prediction for Better Speech Systems

Abstract

Prediction of morphological forms is a well-studied problem and can lead to better speech systems either directly by rescoring models for correcting morphology, or indirectly by more accurate dialog systems with improved natural language generation and understanding. This includes both lemmatization, i.e. deriving the lemma or root word from a given surface form as well as morphological inflection, i.e. deriving surface forms from the lemma. We train and evaluate various languageagnostic end-to-end neural sequence-to-sequence models for these tasks and compare their effectiveness. We further augment our models with pronunciation information which is typically available in speech systems to further improve the accuracies of the same tasks. We present the results across both morphologically modest and rich languages to show robustness of our approach.

Enhancing Sequence-to-Sequence Text-to-Speech with Morphology

“…Moreover in [16], a Bi-LSTM with grapheme input predicted morphological boundaries with an accuracy of 91.1% for Uniysn and 93.8% for Combilex. Similar results were found in other languages in [30]. We intend to evaluate the difference between using predicted and oracle morphological boundaries as input to neural S2S TTS systems in future work.…”

Improved pronunciation prediction accuracy using morphology

“…Morphology prediction is of independent interest and has applications in natural language generation as well as understanding. The problems of lemmatization and morphological inflection have been studied in both contextual (in a sentence, which involves morphosyntactics) and isolated settings (Cohen and Smith, 2007;Faruqui et al, 2015;Cotterell et al, 2016;Sharma et al, 2019).…”

“…G2P fits nicely in the well-studied sequence to sequence learning paradigms (Sutskever et al, 2014), here we use extensions that can handle supplementary inputs in order to inject the morphological information. Our techniques are similar to Sharma et al (2019), although the goal there is to lemmatize or inflect more accurately using pronunciations. Taylor and Richmond (2020) consider improving neural G2P quality using morphology, our work differs in two respects.…”

Proceedings of the 18th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphology

“…Morphology prediction is of independent interest and has applications in natural language generation as well as understanding. The problems of lemmatization and morphological inflection have been studied in both contextual (in a sentence, which involves morphosyntactics) and isolated settings (Cohen and Smith, 2007;Faruqui et al, 2015;Sharma et al, 2019).…”

“…G2P fits nicely in the well-studied sequence to sequence learning paradigms , here we use extensions that can handle supplementary inputs in order to inject the morphological information. Our techniques are similar to Sharma et al (2019), although the goal there is to lemmatize or inflect more accurately using pronunciations. Taylor and Richmond (2020) consider improving neural G2P quality using morphology, our work differs in two respects.…”