Improved speech recognition in noise
作者： B. Landell , R. Wohlford , L. Bahler
DOI: 10.1109/ICASSP.1986.1169211
关键词: Connected speech 、 Voice activity detection 、 Intelligibility (communication) 、 Speech processing 、 Acoustic model 、 Speech recognition 、 Pattern recognition 、 Linear predictive coding 、 Noise 、 Artificial intelligence 、 Noise measurement 、 Computer science 、 Vocabulary 、 Speech enhancement 、 Background noise
摘要: Recent enhancements to a connected speech recognition algorithm have led substantial improvements in accuracy especially for noisy environments. The performance are direct result of three features this algorithm: elimination priori endpoint detection; adaptation templates additive noise; and scoring using likelihood ratio. Experimental results provided which illustrate the both within-vocabulary substitution out-of-vocabulary rejection.
sci-hub.se 参考文章(1)
P. Brown, J. Spohrer, P. Hochschild, J. Baker, Partial traceback and dynamic programming international conference on acoustics, speech, and signal processing. ,vol. 7, pp. 1629- 1632 ,(1982) , 10.1109/ICASSP.1982.1171441