A Holistic Review of Automatic Speech Recognition Systems for Real-time Implementation

Abstract

Once considered a science-fiction idea, Automatic Speech Recognition (ASR) has now become a crucial part of information and communication technology, despite being challenged by various execution-related issues. Over time, ASR has undergone significant changes, evolving from merely responding to sounds to effectively recognizing natural language. With advancements made by researchers, ASR technology and spoken language systems continue to improve. However, there are still challenges in developing flexible solutions that can satisfy user needs in certain situations. In this article, we examine recent techniques related to ASR systems, including pre-processing techniques, classification techniques, feature extraction techniques, and speech recognition techniques.Here some of the recent techniques related to ASR system based on a few factors which are having different application and techniques used for classification. The different application are applicable to the speaker depedent and speaker independent as training part few of them applications are use for real time application and few of them are work on offline basis. One critical aspect of ASR is feature extraction, which involves identifying relevant acoustic features from speech signals to support accurate recognition. To enhance ASR performance, researchers have explored various techniques, including Support Vector Machine (SVM) and Convolutional Neural Network (CNN) for classification. These methods have been used to improve the accuracy of speech recognition and have been applied in numerous applications.