A lot of contributions were dedicated by Speech processing scientists and researchers throughout the globe for English and non-English spoken languages. But very scanty research in Oriya spoken language is done so far. This research work is completely a novel contribution towards the Oriya language which is one of the recognized Indian official languages. Multiple aspects are taken into account in this thesis. Statistical pattern recognition algorithms are utilized in the development of speaker identification and Oriya speech recognition. Speaker identification task is based on the Support Vector Machine (SVM) algorithm whereas Hidden Markov Model (HMM) algorithm is used in Oriya Speech Recognition engine development. Other effective techniques like Dynamic Time Warping (DTW), ontology based back propagation feed forward neural network (BPFF), k-neighborhood etc. also reviewed and analyzed from the point of view of speaker identification and Oriya speech recognition performance. A speech corpus is the integral unit of any speech processing application. Due to the unavailability of public domain Oriya speech corpus, we have established two different speech corpora in Oriya language.