In deep learning applications for Carnatic music (a form of Indian classical music (ICM)), identifying the tonic pitch, also known as Shruti, is a foundational step for tackling more complex tasks such as raga classification and melodic analysis. Musicians construct melodies around the tonic pitch, with instruments tuned accordingly. Due to sudden oscillations in frequencies (Gamakha), the task of tonic identification is made much more complex while working with Carnatic music data. The proposed work treats tonic pitch identification as a classification problem, mapping the predicted tonic frequency to the closest pitch in the Carnatic music scale (A–G#). Rather than relying on a set template, the proposed work implements a novel method with an end-to-end deep learning model that autonomously discovers rules for tonic identification from Mel frequency cepstral coefficients, chroma and spectral features. The Comp Music dataset for Carnatic music was employed in the proposed work, achieving an accuracy rate of 98.76% in tonic identification. This breakthrough underscores the potential of leveraging advanced computational techniques to unravel the complexities embedded within the rich tapestry of Carnatic music, paving the way for deeper insights and enhanced musical appreciation.