Dr Nazimunisa, Gali Lakshmi Pras Anna, Thalla Shiva Kumar, Mohammad Saniya, Mohammad Junaid 03 06 Accent recognition has emerged as an important research area in speech processing due to its wide range of applications in automatic speech recognition ASR speaker identification multilingual virtual assistants language learning systems and humancomputer interaction Variations in pronunciation intonation rhythm and phonetic patterns across different geographical regions often reduce the performance of conventional speech recognition systems Recent advances in machine learning and deep learning have significantly improved accent recognition by automatically extracting complex acoustic and linguistic features from speech signals This paper presents a comprehensive framework for accurate accent recognition by integrating advanced feature extraction techniques with machine learning and deep learning models The proposed framework employs MelFrequency Cepstral Coefficients MFCCs spectrogram analysis and audio preprocessing techniques to extract discriminative speech features Machine learning algorithms such as Support Vector Machine SVM and Random Forest RF are compared with deep learning architectures including Convolutional Neural Networks CNNs Long ShortTerm Memory LSTM networks and hybrid CNNLSTM models Experimental evaluation demonstrates that deep learning models outperform conventional machine learning methods by achieving higher classification accuracy robustness to speaker variability and improved generalization across multiple accents The proposed framework contributes to the development of intelligent speech processing systems capable of recognizing diverse accents with high reliability thereby enhancing speechenabled applications in multilingual and global communication environments Copyright (c) Journal of Science Engineering Technology and Management Science. All rights reserved 2026