Entropy-based algorithm for speech recognition in noisy environment

This paper presents an entropy-based algorithm for endpoint detection and speech recognition in noisy environments. Entropy was developed to identify the speech segments accurately. Results show that this entropy-based algorithm outperforms the energy-based algorithms in putting the endpoints of speech nearly to its real value.