METHOD AND DEVICE FOR ESTIMATING SIGNAL PARAMETERS BY MEANS OF KALMAN FILTRATION IN ORDER TO FREE A LANGUAGE SIGNAL FROM NOISE

摘要

The aim of the invention is to estimate language signals in surroundings that are affected by noise, e.g. during language communication via channels of telecommunications engineering in mobile radio telephone systems, in such a way that the language signal is not incorrectly coded and is not transmitted together with optionally occurring interference noises. This would be particularly unwelcome during language communication in cars via channels of telecommunications engineering in mobile radio systems. Incorrect coding would have negative effects on the quality of the language signal, especially at low bit rates. According to the inventive procedure, the language signal estimation in conditions of noise is based on the algorithm of the Kalman filter. Parameters of the short-time predictor (STP parameters) and parameters of the long-time predictor (LTP parameters) are used in order to optimally estimate the language signal. The short-time and the long-time predictor are known by the model of the language synthesis. Noise parameters are used. The parameters mentioned are used in a representation of the signal in a state space.