A Lyric-Based Approach for Brazilian Music Knowledge Discovery: Brazilian Country Music as a Case Study
Computational techniques can be used to identify musical trends and patterns, helping people filtering and selecting music according to their preferences. In this scenario, researches claim that the future of music permeates artificial intelligence, which will play the role of composing music that best fits the tastes of consumers. So, extracting patterns from this data is critical and can contribute to the music industry ecosystem. These techniques are well known in the field of Musical Information Retrieval. They consist of the audio characteristics extraction (content) or lyrics (context), being the latter preferable because it demands lower computational cost and presenting better results. However, when observing the state of the art, it was found that there is a lack of antecedents that investigate the extraction of Brazilian music patterns through lyrics. In this sense, the main goal of this work is to fill this gap through text mining techniques, analyzing the songs classification in the subgenres of Brazilian country music. This analysis is based on lyrics and knowledge extraction to explain how subgenres differ.
READ FULL TEXT