2007 IEEE International Conference on Granular Computing (GRC 2007) (2007)
San Jose, California
Nov. 2, 2007 to Nov. 4, 2007
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/GrC.2007.95
Speech endpoint detection is an important step in the field of speech analysis, speech synthesis and speech recognition. This paper proposed an endpoint detection algorithm, which used amplitude entropy, spectral entropy and frame energy as feature parameters and utilized RBF neural network as a feature classification system. 170 sentences are used as testing data to detect speech endpoint, which length is from 4 second to 7 second. The experiments show that the testing results using RBF neural network are better than that using entropy alone or BP neural network based on entropy. Keywords--amplitude entropy, neural network, speech endpoint detection, spectral entropy
F. Qiao, X. Zhang and G. Li, "A Speech Endpoint Detection Algorithm Based on Entropy and RBF Neural Network," 2007 IEEE International Conference on Granular Computing (GRC 2007)(GRC), San Jose, California, 2007, pp. 506.