<p>Music is a divine means of expressing what is most beautiful in this world. The majority of people are aware of the differences between Indian and Western classical music. Raga classification plays a crucial function in music information retrieval (MIR) for comprehending the foundations of Indian classical music, in addition to a variety of other responsibilities, including music recommendation systems and database structuring of music files. The work uses a range of methods like pre-processed using spectral centroid-based normalization, pre-emphasis, and silence removal. These techniques enhance the audio quality by normalizing, boosting the amplitude of the signal, and removing the silent portion from the signals. Next, spectral bandwidth, spectral roll-off, zero crossing rate, and Mel-Frequency Cepstral Coefficients (MFCC) are used to extract features. These techniques extract several characteristics regarding pitches, tempo variations, and the performer’s tonic pitch. Based on the Meerkat Optimization Algorithm (MOA), the optimal features are selected to improve the classification accuracy. The refined Bidirectional Long Short Term Memory –Extreme Gradient Boosting (BiLSTM-XGBoost) models are used to train and evaluate these particular features for the task of Raga Identification from a Carnatic Classical Instrumental audio. Compared to the single model, the hybrid model can reduce the risk of overfitting and increase model generalization. Each model can address the weaknesses of others, leading to more reliable predictions. The proposed fine-tuned BiLSTM-XGBOOST approach has an accuracy of 97.3%, recall of 89%, NPV of 98%, and Fall-out ratio of 1.9%. The procedure was applied to the Ragas of Carnatic Classical music, which is a subgenre of Indian classical music.</p>