Multimodal Sentiment Analysis of Nursery Rhymes for Behavior Improvement of Children

Authors

Makarand Velankar, Assistant Professor
MKSSS Cummins College of Engineering Pune and Ph.D. Research Scholar PICT, Pune Maharashtra, India.
Vaibhav Khatavkar, Ph.D. Research Scholar
College of Engineering Pune, Maharashtra, India.
Dr. Parag Kulkarni, CEO 
Chief Scientist at Kvinna Limited, Pune, Maharashtra, India.

Abstract

Introvert children have behavioral issues such as self-centric, poor communication, low confidence, less participation and physically weak, whereas hyperactive children have issues related to non-obedient, aggression, lack of patience, lower concentration levels, and disturbance to others. There is a need to do some remedial measures to improve the behavior of children for their better future. Music interactions with the specific sentiment, i.e. positive sentiments for introvert and negative sentiments for hyperactive children are a proposed approach used. Multimodal analysis of 50 nursery rhymes using cognitive music models with analysis of lyrics and acoustic parameters was performed to select rhymes in specific categories. Specific sets of rhymes were repeatedly played and the behavior was observed and noted for specific predefined parameters by teachers and parents. The improvement noticed in the initial 3 months with about 10 musical sessions per month was in the range of 20 to 40 %. This is probably the first kind of study to use multimodal music sentiment analysis for the behavioral improvement of kids with hyperactive and introvert characteristics. The study also provides effective use of selected acoustic and text features with a normalization approach for sentiment analysis.