Hidden Markov Model for speech recognition system—a pilot study and a Naive approach for speech-to-text model

Abstract

Today’s advancement in the research field has brought a new horizon to design the state-of-the-art systems that produce sound utterance. In order to attain a higher level of speech understanding potentiality, it is of utmost importance to achieve good efficiency. Speech-to-Text (STT) or voice recognition system is an efficacious approach that aims at recognizing speech and allows the conversion of the human voice into the text. By this, an interface between the human and the computer is created. In this direction, this paper introduces a novel approach to convert STT by using Hidden Markov Model (HMM). HMM along with other techniques such as Mel-Frequency Cepstral Coefficients (MFCCs), Decision trees, Support Vector Machine (SVM) is used to ascertain the speakers’ utterances and catalyse these utterances into quantization features by evaluating the likelihood extremity of the spoken word. The accuracy of the proposed architecture is studied, which is found to be better than the existing methodologies.