Preprocessing of Continuous Bengali Speech for Feature Extraction

Abstract

As Voice is the most suitable form of communication, voice-based applications are playing a vital role in modern technology for the last few decades. It is not only the trend of modern and efficient technology but also a new shift of information and technology paradigm. Several research works have been completed on voice-based applications because it has more practical application than any other form of communication. For almost every application, voice signals need to be preprocessed before using it as the input signal. Preprocessing of any data improves the performance of applications. A preprocessing method for feature extraction of continuous Bangla voice has been proposed in this paper. In this method the signal is preprocessed in several steps. In the first step, the noise has been reduced from the signal. Then after balancing the frequency applying pre-emphasis, framing is applied which splits the whole signal into some frames. After that a hamming window and normalization are applied to improve the spectrum and SNR of the preprocessed signal. By following these steps, a clear voice signal, free from noise and unnecessary frequencies has been retrieved.