Speech analysis for alphabets in Bangla language: automatic speech recognition

摘要

This paper presents a technique for recognizing spoken letter in Bengali Language. We first derive feature from spoken letter. Mel-frequency cepstral coefficient (MFCC) has been used to characterize a feature. Dynamic time warping (DTW) employed to calculate the distance of an unknown letter with the stored ones. K-nearest neighbors (KNN) algorithm is used to improve accuracy in noisy environment.