Speaker identification is a bio-metric method that uses speaker speeches to identify that speaker among many speakers. In simulation work, many methods are available for the automatic speaker identification. Nowadays, Siri in iPhone, Google assistance in Google, and almost all android system use bio-metric encryption such as finger-print or speech as a mean of device access. These methods give a better result in clean conditions. However, their performances are reduced under noisy conditions. Most of the recent works are related to develop a noise-robust speaker identification system in computer-based simulation. But, my concern, how to implement an existing speaker identification such as Mel-frequency Cepstral Coefficient-based system in a hardware-level that will be running with low power consumption, but giving the human-level performance.