HomeReseach Talks ➤ 091 15 05 2024

Analyzing Emotional Responses through Sinhala Comments on Sinhala Music: A Dataset Study

Yomal De Mel
Slides Video

This research delves into the realm of Music Information Retrieval (MIR) and Music Emotion Recognition (MER) within the context of Sinhala songs, an area that has been relatively underexplored in the broader landscape of music research. By leveraging social media comments as primary data sources, the study aims to unravel the emotional dimensions embedded within diverse Sinhala musical traditions. The dataset comprises comments extracted from YouTube videos featuring 20 Sinhala songs, meticulously curated to ensure linguistic fidelity and relevance. Through a combination of linguistic and computational methodologies, the research explores the emotional responses evoked by Sinhala music, with a focus on valence and arousal dimensions. The findings not only contribute to the understanding of music emotions within the Sinhala cultural environment but also demonstrate the efficacy of merging cultural insights with machine learning techniques for nuanced emotional analysis. This paper analyzed 27 YouTube videos related to 20 Sinhala songs, extracting a total of 93,116 comments. Through the use of advanced filtering methods and transliteration mechanisms, the study was able to refine the dataset, extracting 63,471 Sinhala comments suitable for emotion analysis in the Sinhala language domain. Additionally, 379 stop-words specific to the Sinhala language were algorithmically derived, with 98 of them matching against the NLTK corpus of English stop words once translated to English. By combining linguistic and computational approaches, this research provides a comprehensive analysis of the emotional landscape of Sinhala music, shedding light on previously unexplored aspects of music emotions within the Sinhala cultural context. The meticulously curated dataset and the derived stop words contribute valuable resources for further research in the fields of MIR and MER, highlighting the potential of computational techniques in unravelling the complexities of musical experiences within diverse cultural traditions.

    Page: /