| 摘要 |
In general terms the present invention is a method for the coded transmission of voice signals. A voice signal is digitized and supplied to an encoder and the estimated input signal is subtracted from the digitized voice signal. The remaining remainder signal is normed in that it is divided by its estimated intensity. The norm signal which has arisen as a result thereof is supplied to a quantizer/coder. The output signal thereof is transmitted via a channel and is supplied to a decoder and is also subsequently supplied to an inverse quantizer whose output signal is multiplied by the estimated intensity. The signal that proceeded from the multiplication of the estimated intensity by the output signal of the inverse quantizer is added to the estimated signal, as a result whereof the control signal for a predictor is formed which generates the estimated signal and the decoding ensues analogously to the coding. The estimated intensity is calculated in the following way: the amount of the output signal of the inverse quantizer is formed, and then deviations of the amount of this output signal from 1 are accumulated and limited upon attenuation of the higher-frequency signal parts. The calculation of the estimated intensity ensues using the exponential function 2x, whereby only the whole-numbered part of x is taken into consideration. |
