This paper compares the speech coder and speaker recognizer applications, showing some parallelism between them. In this paper, some approaches used for speaker recognition are applied to speech coding in order to improve the prediction accuracy. Experimental results show an improvement in Segmental SNR (SEGSNR).