Introduction - Robust Emotion Recognition Using Spectral and Prosodic Features - page 11

1.9.2 Emotion Recognition using Global and Local Prosodic
Features Extracted from Words and Syllables

Most of the existing emotion recognition studies have focused on global prosodic
features. In this topic, local (dynamic) prosodic features have been proposed for
emotion recognition. The combination of global and local prosodic features is also
used for discriminating the emotions. The contribution of different speech segments
like syllables, words and sentences and their positions (initial, middle, and final) are
also studied in detail, in the context of emotion recognition using prosodic features.

1.9.3 Emotion Recognition using Combination of Features

In this topic, we have explored emotion-specific spectral and prosodic features for dis-
criminating the emotions. The proposed features may provide some complementary
or supplementary information with respect to each other. Therefore, for exploiting
the complementary emotion-specific information provided by individual features,
appropriate fusion techniques are proposed in this topic for combining the evidences
provided by various features.

1.9.4 Emotion Recognition on Real Life Emotional Speech
Database

The proposed spectral and prosodic features are evaluated on real life emotional
speech database. In this work, a Hindi movie database is used to represent real
life emotions. The recognition performance using proposed features seems to be
better compared to conventional spectral and prosodic features. ATwo-stage emotion
recognition approach based on speaking rate has been proposed for improving the
emotion recognition performance further.

1.10 Organization of the Topic

The evolution of ideas presented in this topic is given at the end of the chapter in
Table 1.2 . The chapter-wise organization of the topic is given below.

￿ Chapter 1 : The Introduction discusses about the characterization of emotions

by machines and human beings. The philosophy of emotions with respect to the
psychology and speech production aspects is briefly discussed. Different types of
emotional speech corpora, contemporary issues in speech emotion recognition,
applications, objectives and scope of this topic are briefly discussed. Evolution of
the ideas of this topic is given at the end of the chapter.