Vocal and Text Emotion Formant Analysis with Graph Lasso and Empirical Mode Decomposition in Political Speech

Posted: 13 Feb 2023

See all articles by Marta Campi

Marta Campi

Institut Pasteur - Hearing Institute

Ioannis Chalkiadakis

Institut des Systèmes Complexes de Paris Île-de-France / CNRS - UAR 3611

Gareth Peters

University of California Santa Barbara; University of California, Santa Barbara

Date Written: February 10, 2023

Abstract

In this work, we aim to identify sentiment lying at different speech time scales and frequency components as well as their corresponding pronounced text components. This goal is achieved through the construction of text and speech indicators by relying on the signal processing method known as the Empirical Mode Decomposition. This corresponds to a non-stationary decomposition technique able to isolate oscillation modes of a given signal carrying individual frequency components of the original signal. The framework consists of comparing political speeches to robot-generated voices using standard Artificial Intelligence (AI) reading tools which, in practice, have to address several challenges to reproduce human life emotions.

Keywords: EMD, Graph Lasso, Sentiment Analysis, Text Features

Suggested Citation

Campi, Marta and Chalkiadakis, Ioannis and Peters, Gareth, Vocal and Text Emotion Formant Analysis with Graph Lasso and Empirical Mode Decomposition in Political Speech (February 10, 2023). Available at SSRN: https://ssrn.com/abstract=4353272

Marta Campi (Contact Author)

Institut Pasteur - Hearing Institute ( email )

France

Ioannis Chalkiadakis

Institut des Systèmes Complexes de Paris Île-de-France / CNRS - UAR 3611 ( email )

113 Rue Nationale
Paris, 75013
France

HOME PAGE: http://www.iscpif.fr/

Gareth Peters

University of California Santa Barbara ( email )

Santa Barbara, CA 93106
United States

University of California, Santa Barbara ( email )

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Abstract Views
104
PlumX Metrics