{"citation":{"mla":"Haeb-Umbach, Reinhold, and Hermann Ney. “Improvements in Beam Search for 10000-Word Continuous-Speech Recognition.” IEEE Transactions on Speech and Audio Processing, 1994.","bibtex":"@article{Haeb-Umbach_Ney_1994, title={Improvements in beam search for 10000-word continuous-speech recognition}, journal={IEEE Transactions on Speech and Audio Processing}, author={Haeb-Umbach, Reinhold and Ney, Hermann}, year={1994} }","ama":"Haeb-Umbach R, Ney H. Improvements in beam search for 10000-word continuous-speech recognition. IEEE Transactions on Speech and Audio Processing. 1994.","apa":"Haeb-Umbach, R., & Ney, H. (1994). Improvements in beam search for 10000-word continuous-speech recognition. IEEE Transactions on Speech and Audio Processing.","chicago":"Haeb-Umbach, Reinhold, and Hermann Ney. “Improvements in Beam Search for 10000-Word Continuous-Speech Recognition.” IEEE Transactions on Speech and Audio Processing, 1994.","ieee":"R. Haeb-Umbach and H. Ney, “Improvements in beam search for 10000-word continuous-speech recognition,” IEEE Transactions on Speech and Audio Processing, 1994.","short":"R. Haeb-Umbach, H. Ney, IEEE Transactions on Speech and Audio Processing (1994)."},"year":"1994","type":"journal_article","language":[{"iso":"eng"}],"date_updated":"2022-01-06T06:51:08Z","_id":"11796","status":"public","date_created":"2019-07-12T05:28:25Z","author":[{"last_name":"Haeb-Umbach","id":"242","first_name":"Reinhold","full_name":"Haeb-Umbach, Reinhold"},{"last_name":"Ney","first_name":"Hermann","full_name":"Ney, Hermann"}],"publication":"IEEE Transactions on Speech and Audio Processing","department":[{"_id":"54"}],"title":"Improvements in beam search for 10000-word continuous-speech recognition","user_id":"44006","abstract":[{"text":"The authors describe the improvements in a time-synchronous beam search strategy for a 10000-word continuous-speech recognition task. Basically they introduced two measures, namely a tree organization of the pronunciation lexicon and a novel look-ahead technique at the phoneme level. The experimental tests performed showed that the number of state hypotheses could be reduced from 50000 to 3000, i.e., by a factor of about 17. At the same time, the word error rate did not increase.","lang":"eng"}]}