This email address is being protected from spambots. You need JavaScript enabled to view it.
 
+7 (4912) 72-03-73
 
Интернет-портал РГРТУ: http://rsreu.ru

UDC 621.396.669

NEURAL NETWORKS PREDICTOR SPEECH SIGNALS

S. N. Kirillov, PhD (technical sciences), full professor, Head of the Department, RSREU, Ryazan; This email address is being protected from spambots. You need JavaScript enabled to view it.
E. S. Sazonova, master, This email address is being protected from spambots. You need JavaScript enabled to view it.

The theoretical and practical aspects of design and optimization structures of neural network implementations predictors of speech signals are offered. The aim of the work is to justify bath implementation advantages predictors based on artificial neural networks compared with known predictors based on nonrecursive FIR filters. The possibility to reduce the prediction order from 10% to 60% at the same error, and the prediction error reduction from 15% to 70% at the same manner is proved. The increase of subjective evaluation of speech signal quality on 0,1-0,45 points according to MOS scale is achieved which can be viewed as a considerable advantage of given systems.

Keywords: artificial neural network, perceptron, linear regression network, FIR filter, predictor, speech signals.

 Скачать статью