In this paper we propose a new parameterization algorithm based on nonlinear prediction, which is an extension of the classical LPC parameters. The parameters performances are estimated by two different methods: the Arithmetic-Harmonic Sphericity (AHS) and the Auto-Regressive Vector Model (ARVM). Two different methods are proposed for the parameterization based on the Neural Predictive Coding (NPC): classical neural networks initialization and linear initialization. We applied these two parameters to speaker identification. The fist parameters obtained smaller rates. We show for the first parameters how they can be combined with the classical parameters (LPCC, MFCC, etc.) in order to improve the results of only one classical parameterization (MFCC provides 97.55% and MFCC+NPC 98.78%). For the linear initialization, we obtain 100% which is great improvement. This study opens a new way towards different parameterization schemes that offer better accuracy on speaker recognition tasks.