Investigations on joint-multigram models for grapheme-to-phoneme conversion
Investigations on joint-multigram models for grapheme-to-phoneme conversion
复制标题
字素到音素转换的联合多语法模型的研究
DOI：
10.21437/icslp.2002-78
复制
发表时间：
2002
期刊：
影响因子：
--
通讯作者：
H. Ney
中科院分区：
文献类型：
--
作者：
M. Bisani;H. Ney
我们提出了一种完全由数据驱动、独立于语言的方法来构建字素到音素转换器。我们将联合多重图方法应用于比对问题，并使用标准的languagemodellingtechniquestomodeltranscriptionprobabilities.最优模型参数、训练步骤和corpussizeindetail.ExperimentswereconductedonEnglish andGermanpronunciationlexica.Ourproposedtrainingscheme performsbetterthanpreviouslypublishedones.Phonemeerror率效果均为3。英语为98%，0。德语达到了51%。
We present a fully data-driven, language independent way of building a grapheme-to-phoneme converter. We apply the joint-multigram approach to the alignment problem and use standard languagemodellingtechniquestomodeltranscriptionprobabilities. Westudymodelparameters,trainingproceduresandeffectsof corpussizeindetail.ExperimentswereconductedonEnglish andGermanpronunciationlexica.Ourproposedtrainingscheme performsbetterthanpreviouslypublishedones.Phonemeerror ratesaslowas 3 . 98% for English and 0 . 51% for German were achieved.