What makes a song hit the Thailand Top 100 Chart? A team at CMKL University in the (Entertainment Innovation Center) EIC Program discovers all the essential factors that drive songs to popularity and successfully utilizes AI to support artists and musicians in making music that appeals to their listeners while giving them creative freedom.
The idea for AI Music Creation was born from the shared interests and experience of all the team members: Gunyootapong Nopakun (Barge) was a DJ for 15 years, Pongsiri Hetrakul (Tom) was a member of world-class Mor Lam Band, 'Bangkok Paradise,' and Jakarin Sirikulthorn is deep into Data and Machine Learning. They've all seen trends come and go, and Thai people's music tastes evolve. Especially with streaming platforms like Youtube coming into play, it gives people more access to discover music they like. With this, the team sought to work in Data Science, Machine Learning, and Design Thinking to create an AI that predicts trends and success factors of music just like the minds of humans.
Working together in a simple workflow, these Researchers use Python to extract important data from the Spotify API, such as audio features and streaming volumes of the Thailand Top 200 chart from the past 5 years. This data is then clustered to calculate the music preferences of listeners - what do these hit songs have in common? It is vital to confirm these findings with key figures in the industry, from producers to musicians, to see if it's what they need as human-centric findings are key. The last step is using Natural Language Processing (NLP) to dig into the song's actual content to gain more in-depth data.
The team is on track to move towards their ambitious goal where the deliverables aim to cluster music genres from the data and discover what producers and artists want in terms of aspects that influence songs, such as keywords or TikTok popularity.
The team has witnessed the struggles of the music industry and understands that a song's popularity is influenced by both art and business. In the age of data, we have all the information we need to support artists create music that appeals to their specific audiences and help everyone's music become what they want it to be.
.png)
