Automatic identification of Malvani dialects from audio signal based on hybrid FFO-TSO with deep neural network

Marathi language is originated from the geographical regions of Maharashtra, which include several different varieties of Malvani. The identification of Malvani dialects has been suggested because the analysis method for those Malvani languages is a little challenging for identifying itself. Malvani dialects have been identified in a number of ways, all of which have some negative impact on the accuracy of the results and the requisite level of precision and mistake occurrence. So as to eradicate the above-mentioned issues, automatic identification of Malvani dialects is proposed based on hybrid optimization with Multilayer Perceptron (MLP) and Deep Neural Network (DNN). The audio signal related to Marathi and Malvani are gathered and pre-processed by employing a pre-emphasis filter, adaptive thresholding, framing and windowing. Then, the features are extracted from the pre-processed signal using the Fast Fourier Transform (FFT) and the Mel Frequency Cepstral Coefficient (MFCC). The appropriate features are chosen using a hybrid Fennec Fox (FFO) and the Tuna Swarm Optimization (TSO). Finally, the chosen features are classified using MLP and DNN classifiers. The appropriate signals for determining whether audio is Marathi or Malvani will be determined based on the best results from the MLP and DNN. According to the results of the experimental research, the proposed MLP technique achieves 95% accuracy, 94% precision and DNN achieves 96% accuracy, 96% precision. Thus, language detection using audio signals made easier using this deep learning based automated model.