Learning a Lexicon and Translation Model from Phoneme Lattices

Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 2377–2382, Austin, Texas, November 1-5, 2016. c⃝2016 Association for Computational Linguistics Learning a Lexicon and Translation Model from Phoneme Lattices Oliver Adams,♠♥ Graham Neubig,♣♥ Trevor Cohn,♠ Steven Bird,♠ Quoc Truong Do,♥ Satoshi Nakamura♥ ♠The University of Melbourne, Australia ♣Carnegie Mellon University, Pittsburgh, PA, USA ♥Nara Institute of Science and Technology, Japan Abstract Language documentation begins by gather- ing speech. Manual or automatic transcrip- tion at the word level is typically not possi- ble because of the absence of an orthography or prior lexicon, and though manual phone- mic transcription is possible, it is prohibitively slow. On the other hand, translations of the minority language into a major language are more easily acquired. We propose a method to harness such translations to improve auto- matic phoneme recognition. The method as- sumes no prior lexicon or translation model, instead learning them from phoneme lattices and translations of the speech being tran- scribed. Experiments demonstrate phoneme error rate improvements against two baselines and the model’s ability to learn useful bilin- gual lexical entries. 1 Introduction Most of the world’s languages are dying out and have little recorded data or linguistic documentation (Austin and Sallabank, 2011). It is important to ad- equately document languages while they are alive so that they may be investigated in the future. Lan- guage documentation traditionally involves one-on- one elicitation of speech from native speakers in or- der to produce lexicons and grammars that describe the language. However, this does not scale: lin- guists must first transcribe the speech phonemically as most of these languages have no standardized orthography. This is a critical bottleneck since it takes a trained linguist about 1 hour to transcribe the phonemes of 1 minute of speech (Do et al., 2014). Smartphone apps for rapid collection of bilin- gual data have been increasingly investigated (De Vries et al., 2011; De Vries et al., 2014; Reiman, 2010; Bird et al., 2014; Blachon et al., 2016). It is common for these apps to collect speech segments paired with spoken translations in another language, making spoken translations quicker to obtain than phonemic transcriptions. We present a method to improve automatic phoneme transcription by harnessing such bilingual data to learn a lexicon and translation model directly from source phoneme lattices and their written tar- get translations, assuming that the target side is a major language that can be efficiently transcribed.1 A Bayesian non-parametric model expressed with a weighted finite-state transducer (WFST) framework represents the joint distribution of source acoustic features, phonemes and latent source words given the target words. Sampling of alignments is used to learn source words and their target translations, which are then used to improve transcription of the source audio they were learnt from. Importantly, the model assumes no prior lexicon or translation model. This method builds on work on phoneme transla- tion modeling (Besacier et al., 2006; St¨uker et al., 2009; Stahlberg et al., 2012; Stahlberg et al., 2014; Adams et …

Content truncated to 500 words