Published Paper Details:

-- In Indian context due to cultural diversity and difference languages in use, it's very of important to understand the language being spoken. It becomes even important when different spoken language team's come together for common goal. To handle this challenge, Spoken Language Identification (SLID) helps to automatically determine the language without having any background of the specific language. This applies even for foreign languages. people from different countries are coming together virtually or face to face. In India and global scenario identification becomes very challenging due to linguistic diversity, diverse phonetics, un developed or limited used language, dialectal variations and code mixing. This study provides more insights on the various major approaches studied, followed and ongoing development on SLID Mainly in SLID classical machine learning methods such as Support Vector Machines (SVM), K-Nearest Neighbors (KNN), Random Forests (RF), Extreme Learning Machines (ELM) and contemporary deep learning methods, including Convolutional Neural Networks (CNN), Recurrent Neural Networks with Long Short-Term Memory (RNN-LSTM), Deep Neural Networks (DNN), Convolutional Recurrent Networks (CRNN), and ensemble learning architectures are used. We intend to analyze trends in feature extraction, including MFCC, Melspectrogram, i-vectors and self-supervised embedding like wav2vec. Challenges in real-world SLID, such as short utterances, noise robustness, and multilingual speech datasets. The proposed workflow integrates preprocessing, feature extraction, DL training, and classification, demonstrating potential for robust SLID in multilingual and noisy environments.