Feature Extraction techniques for Classification of Emotions in Speech Signals

Automatic speech emotion recognition is a process of recognizing emotions in speech. This has wide applications in the area of phsycatrics help and in robotics'he human computer interaction the challenging area of research. Any effective HCI system has two sections Training and testing. The techniques used in the system are feature extraction and classification. This paper focuses on the brief introduction of the GFCC feature extraction, optimization algorithm and the back propagation neural network for the classification of the emotions in speech. Speaker recognition refers to recognize the person from their speech. The speech signal contains the message being spoken, the emotional state of the speaker and the information of the speaker. So the speech signal can be used for the recognition of the speaker and the emotional state of the speaker .Emotion recognition in speech extracts the speech features in each utterance. It is the process of automatically recognizing who is speaking and in which emotional state the words are spoken on the basis of features present in the speech signal. (1).Speaker recognition can be text independent and text dependent .The context dependent has more accuracy. Speaker recognition is helpful in the areas such as voice dialling, banking by telephone, telephone shopping, database access services, information services, and security control for confidential information areas voice mail and remote access to computers (2).The detection of emotions in speech is gaining attention in wide range of application like mostly used to develop wide range of application like application for call centre and learning, gaming software, security applications and machine translation. Influence of emotional state of human speech in speaker recognition is very high (3). The term "emotion" can refer to an extremely complex state associated with a wide variety of mental, physiological and physical events. Emotional speech database is valuable for this speaker recognition. In a generalized way, a speech emotion recognition system is an application of speech processing in which the patterns of derived speech features (MFCC, pitch) are mapped by the classifier (HMM) during the training and testing session using pattern recognition algorithms to detect the emotions from each of their corresponding patterns. The technique is synonymous to speaker recognition system but its different approach to detect emotions makes it intelligent and adds security to achieve better service in various applications (4). Different techniques are used for the feature extraction which is the first step in emotion recognition. These features are the input to the classifier for the classification of the emotions. If the features extracted are chosen carefully it is expected that the features set will extract the relevant information from the input data in order to perform the desired task using this reduced representation instead of the full size input. II. STRUCTURE OF SPEECH RECOGNITION The speech emotion recognition needs to extract short acoustic and prosody"s feature parameters reflecting emotion, and distinguish through a variety of classifier means. The Speech Recognition System can be divided into the following categories as Signal Preprocessing, Feature Extraction, and Speech Classification. The block diagram for ER system is: