Further results on the recognition of a continuously read natural corpus
作者： L. Bahl , R. Bakis , P. Cohen , A. Cole , F. Jelinek
DOI: 10.1109/ICASSP.1980.1170862
关键词:
摘要: Further results have been obtained on the recognition of continuously read sentences from a natural language corpus laser patents. The vocabulary is limited to 1000 most frequently occurring words in corpus. Our model task has perplexity 24.1 (corresponding an entropy 4.6 bits/word). This paper describes modifications and improvements system which resulted lowering word error rate previously reported 33.1% 8.9%.
sci-hub.se 参考文章(4)
L. Bahl, R. Bakis, P. Cohen, A. Cole, F. Jelinek, B. Lewis, R. Mercer, Recognition results for several experimental acoustic processors ICASSP '79. IEEE International Conference on Acoustics, Speech, and Signal Processing. ,vol. 4, pp. 249- 251 ,(1979) , 10.1109/ICASSP.1979.1170736
L. Bahl, J. Baker, P. Cohen, F. Jelinek, B. Lewis, R. Mercer, Recognition of continuously read natural corpus ICASSP '78. IEEE International Conference on Acoustics, Speech, and Signal Processing. ,vol. 3, pp. 422- 424 ,(1978) , 10.1109/ICASSP.1978.1170402
Frederick Jelinek, Continuous speech recognition by statistical methods Proceedings of the IEEE. ,vol. 64, pp. 532- 556 ,(1976) , 10.1109/PROC.1976.10159
F. Jelinek, L. Bahl, R. Mercer, Design of a linguistic statistical decoder for the recognition of continuous speech IEEE Transactions on Information Theory. ,vol. 21, pp. 250- 256 ,(1975) , 10.1109/TIT.1975.1055384