Text Independent Amharic Language Dialect Recognition: A Hybrid Approach of VQ and GM

In Amharic language there are four main different types of dialects these are Gojjam (Gojjamegna), Wollo (Wollogna), Shewa (Shewagna) and Gonder (Gonderegna). In this paper a hybrid approach of VQ(vector quantization) and GMM(Gaussian Mixture Models) have been used for classifying dialects of Amharic language. For our data set a total of 100 speakers for each group of dialects are considered. Mel frequency cepstral coefficients (MFCC) feature vectors are used to recognize the dialects of speakers. To see the effect of the number of these feature vectors on the performance of the system, MFCC, MFCC and MFCC vectors are used. When 25 speakers are considered from areas, 85.9% accuracy achieved. After conducting this experiment, the number of speakers are increased to100, which is the maximum number of dialect speakers for our experiment, 92.7% accuracy achieved for the given dialects.