AFFECTIVE COMPUTING

Introduction to speech emotional recognition

 

The subject of recognising the emotion in speech has become more relevant in the modern world due to the development of natural interfaces of human and machine communication. The ability to recognise emotion in speech accurately would bring about many benefits to the world, one of which is it would result in better responds from machines to human. 

 

Extracting an accurate emotion from a speech is not an easy task. It requires a large set of database which is difficult to acquire as it is difficult to extend databases from other cases due to the inter-speaker differences. However, studies have shown a promising result in classifying six emotions which are least affected by the inter-speaker differences. Anger, Happiness, Sadness, Disgust, Fear , and Surprise.

 

Emotion classification can be done by looking at the Global prosodic features. 

 

 

We can also classify emotion using Local spectral features.

 

 

Or by anaysing the words in the context of the speech.