Conversational Emotion Recognition using Text and Audio Modalities

Kupekar, Raj Ravindra (2020) Conversational Emotion Recognition using Text and Audio Modalities. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (938kB) | Preview
Preview	PDF (Configuration manual) Download (835kB) | Preview

Abstract

In this recent advancement, extraction and identification of human emotions plays a crucial role in developing interpersonal relationship between human and machine. Emotion recognition system are now been adopted in TV industries for training purposes of the performers to improve their acting skills for connecting the audiences. Accordingly, research is been carried out to study the effect of the emotional behaviour of human using various modalities independently. In this research work, a MELD database is been used which is a conversational-based repository originated from a ‘Friends’ TV series. Here, two independent unimodal networks are implemented using the text and audio modalities and their accuracy and performances are relatively compared for any differences. Accordingly, for text unimodal a Bi-LSTM model is observed to be the efficient model with an accuracy of 75% while a LSTM model is seen to be the superior model for audio modality with its highest accuracy of 47%.

Item Type:	Thesis (Masters)
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > QA Mathematics > Computer software T Technology > T Technology (General) > Information Technology > Computer software
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Dan English
Date Deposited:	22 Jan 2021 15:00
Last Modified:	22 Jan 2021 15:00
URI:	https://norma.ncirl.ie/id/eprint/4452

Actions (login required)

View Item