Feature Extraction using Wavelet Transform and Euclidean Distance for speaker recognition system
Arundhati Niwatkar, Yuvraj Krishnarao Kanse · 2020
The current speaker recognition system has many issues. i.e., channel mismatching, noises, emotions, idioms, etc. To build a reliable and robust speaker recognition system, it's essential to implement the precise feature extraction technique. A first major step in speaker recognition system is feature extraction from the speech signal for that speaker. The speaker model can be build using these features, So that one can extract correct and useful features. The traditional method for feature extraction is MFCC. Here, we have proposed a new technique to obtain the features. We are going to use the wavelet transform technique along with Euclidean Distance computational Technique, which gives multidimensional features for the speaker.