Speaker Recognition Improvement for Degraded Human Voice using Modified-MFCC with GMM

Speaker’s audio is one of the unique identities speaker. Nowadays not only humans but machines can also identify by their audio. Machines different properties human voice and classify speaker from speaker’s Speaker recognition still challenging with degraded limited dataset. be identified effectively when feature extraction more accurate. Mel-Frequency Cepstral Coefficient (MFCC) mostly used method for extraction. We are introducing improved effective signal. This article presents experiment results modified MFCC Gaussian Mixture Model (GMM) on uniquely developed uses signal transforms it into a numerical value characteristics, which utilized to recognize efficiently help data science model. Experiment high background noise comes covers, Sampling Frequency (SF) impacts “Signal Noise Ratio” (SNR) low (up 1dB) in overall identification process. With MFCC, we have observed SNR upto 1dB due SF frequency range mel-scale triangular filter.