Vector Quantization based Speech Recognition System 12+

🔖 Automatic Speech Recognition (ASR) has progressed considerably over the past several decades, but still has not achieved the potential imagined at its very beginning. Almost all of the existing applications of ASR systems are PC based. This work is an attempt to develop a speech recognition system that is independent of any PC support and is small enough in size to be used in a daily use consumer appliance. The proposed system would recognize isolated word utterances from a limited vocabulary, provide speaker independence, require less memory and be cost-efficient compared to present ASR systems. In this system, isolated word recognition is performed by using Vector Quantization (VQ) and Mel-Frequency Cepstral Coefficient (MFCC). The final system has been implemented on a vero board with an ATMEGA32 microcontroller. Learning and recognition algorithm have been used to recognize the speech utterances.