Abstract: Speech processing has always been so important application area of digital signal processing. The various fields are available for researching in speech processing that are speech recognition, speaker recognition, speech synthesis, speech coding etc. The objective of Vachantar Lokbhasha, A Speech to text conversion for Marathi is to recognize speaker’s speech and then convert it to text. This text will be saved to a file and that file can be further used. Automatic speaker recognition is to extract and recognize the information about speaker’s speech. Feature extraction is the first step for this. Many algorithms are stated by the researchers for feature extraction. In this work, the Mel Frequency Cepstral Coefficient (MFCC) feature extraction algorithm has been used for designing this system. In this Artificial Neural Networks will be used for feature classification.

Keywords: Automatic Speech Recognition (ASR), Mel-Frequency Cepstral Coefficients (MFCC), Feature Extraction, Neural Networks.