Review of Multiple Language Classification Using Deep Neural Networks

Abstract

The log-energy parameter, generally derived from a full-band spectrum, could be a essential feature unremarkably employed in automatic speech recognition (ASR) systems. However, log-energy is tough to estimate faithfully within the presence of background. This paper discuss regarding 2 techniques of speech recognition that ar sturdy log-energy estimation and its dynamic change enhancement for in-car speech recognition and recognition of speech in additive and convolutional noise supported adherent spectral process. robust Log-Energy Estimation and its Dynamic change enhancement for In-car Speech Recognition is more practical than the opposite