T¨ urkce Haber B¨ ultenleri icin Dil Modelleme Yaklas¸imlarinin Kars¸ilastirmasi Comparison of Language Modeling Approaches for Turkish Broadcast News
Tuncay Aksungurlu, Siddika Parlak, Haşim Sak · 2008
In this paper, we investigate the performance of several lan- guage modeling approaches on a speech recognition system for Turkish Broadcast News. The agglutinative structure of Turk- ish introduces a high out-of-vocabulary rate and hence increases word error rate. To eliminate out-of-vocabulary problem, we utilize various sub-word models. In addition, we experiment with high vocabulary sizes. Since the models are statistical, we expect an improvement in performance as the amount of train- ing data increases. We build word and sub-word language mod- els using various amounts of corpora and compare their recog- nition performance.