Abstract:
"Advanced music recommendation algorithms have grown more necessary due to the
development of digital music outlets, especially for local music like Tamil songs. The great
range of Tamil music genres and the varied preferences of listeners are frequently ignored by
current systems, resulting in a generic and disappointing user experience. By creating a
customized recommendation engine that matches Tamil songs to users' tastes and moods, this
project aims to solve these limitations.
The study used a hybrid recommendation system method, including audio analysis and song
information available through Spotify's Web API, to address this difficulty. The process
included gathering track metadata, evaluating song characteristics, and applying machine
learning algorithms to categorize songs into mood-based groups. Over time, the system's ability
to learn from user interactions will improve personalization.
The system's capacity to efficiently match songs with user moods has been proven by the
enabling initial findings. For mood categorization, preliminary testing with confusion matrices
and accuracy curves has demonstrated a positive connection between expected and real user
responses."
