• Akanksha Gupta
• Anamika Singh

Music Emotion Recognition (MER) is an important research area within Music Information Retrieval that focuses on automatically identifying emotional characteristics conveyed by music. Although deep learning approaches have shown promising performance, most existing studies rely heavily on vocal or lyrical information and are largely based on Western music datasets. As a result, emotion recognition from instrumental Hindi music remains relatively underexplored. This paper presents a Convolutional Neural Network (CNN)-based framework for emotion recognition using instrumental (voice-removed) Hindi music from the MER500 dataset. Audio signals are converted into spectrogram representations using the Short-Time Fourier Transform and used as input to a CNN for automatic learning of discriminative spectral–temporal features. Experiments are conducted under two configurations: a five-category classification setting (Devotional, Happy, Party, Romantic, and Sad) and a four-category setting obtained by removing the Happy emotion class, which exhibits high ambiguity in instrumental music. The five-category experiment achieves an overall classification accuracy of 64% with a macro-averaged F1-score of 0.64, while excluding the Happy class improves performance to a test accuracy of 67.80% and a macro F1-score of 0.68. The results demonstrate that instrumental Hindi music contains meaningful emotional cues and emphasize the importance of emotion taxonomy design when performing emotion recognition without vocal or lyrical information. This work contributes to culturally diverse MER research and provides a foundation for future multimodal emotion recognition studies.

Music Emotion Recognition, Convolutional Neural Network, Instrumental Hindi Music, Deep Learning

"Deep Learning-Based Emotion Recognition from Instrumental Hindi Music Signal", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 1, page no.b429-b436, January-2026, Available :http://www.jetir.org/papers/JETIR2601162.pdf

"Deep Learning-Based Emotion Recognition from Instrumental Hindi Music Signal", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 1, page no. ppb429-b436, January-2026, Available at : http://www.jetir.org/papers/JETIR2601162.pdf