In this paper, we present a new speech coding algorithm with variable bit rate for telephonic domain. The properties of the Discrete-Time Wavelet Transform and several human psycho-acoustic features are integrated in order to obtain a high quality of coded signal, close to original. This approach allows to reduce the mean bit rate in the range of 10 kbit/s. We describe there the main steps of the coding algorithm. More specifically, we provide the solutions to the following specific problems: choosing the type of the Wavelet Transform, selecting the wavelets coefficients to send, quantifying all sending information (coefficients values and auxiliary) and generating the associated codes.