In the current scenario, speech recognition for several languages is becoming more popular. Recognizing speech is a very difficult task in the Malayalam language. This project aims to establish a Formal Malayalam Speech to Text converter for the language of Malayalam. The system considers only isolated words with constrained vocabulary. The word which is spoken by the speaker is given as the input to the system is presented in the display as the output. We are using deep learning and feature extraction techniques for this project. The proposed system is taking around 5-10 isolated words for tutoring the machine. Since the system is depending on the speaker voice, at the beginning the words are stored in .wav (waveform audio) file for training procedure. Several samples are stored and trained for each word. The input audio word will be collated with these stored words..........
Keywords: Constrained vocabulary, ANN, LSTM, HMM, MFCC, Extraction techniques, Deep learning, Pre-processing, Syllabification.
[1]. C. Kurian and K. Balakrishnan, "Speech recognition of Malayalam numbers", Nature & Biologically Inspired Computing (NaBIC), pp. 1475-1479, 2009.
[2]. Neha Chauhan, Mahesh Chandra, "Speaker recognition and verification using Artificial neural network", IEEE Conference The international Conference on Wireless Communications, Signal Processing and Networking (WiSPNET) pp.1173-1176, 22-24 March, 2017.
[3]. Neha Chauhan, Tsuyoshi Isshiki, Dongju Li," Speaker Recognition Using LPC, MFCC, ZCR Features with ANN and SVM Classifier for Large Input Database", IEEE, ICCCS Singapore, Feb,2019.
[4]. Vimala. C, V. Radha, "Efficient Acoustic Front-End Processing for Tamil Speech Recognition using Modified GFCC Features", International Journal Image, Graphics and Signal Processing, DOI: 10.5815/ijigsp.2016.07.03, 2016, pp.22-31 [5]. Dr. C. Kurian, "Speech database and text corpora for Malayalam language automatic speech recognition technology", Conference of The Oriental Chapter of International Committee for Coordination and Standardization of Speech Databases and Assessment Techniques(O-COCOSDA),2016. doi:10.1109/icsda.2016.7918975