Abstract:In recent years, speech emotion recognition has attracted more and more attention. In this paper, using PCA-based spectral features of the speech signal and HMM, we present a new framework for the text independent recognition of emotions from voice. Contrary to the traditional feature extraction method based on principal components analysis(PCA), the proposed PCA-based spectral features are computed on each set of data belonging to different emotion, rather than on the whole training set, moreover, each emotion subspace is constructed with the eigenvectors corresponding to the smaller eigenvalues rather than the bigger eigenvalues. Then, the proposed features are used to train HMM for realizing the speech emotion recognition. For finding the effectiveness of these features in recognition of different emotions in different languages, two different data-sets are used to evaluate the proposed method: the Danish emotional speech database and the Chinese emotional speech database. For the Danish database, it was shown that higher classification rate can be achieved on the new feature vectors extracted by proposed method. The proposed features, traditional PCA-based spectral features and original spectral features yielded close classification performance for the Chinese database. Copyright ©2009 Binary Information Press.

Discussion on Speech Emotion Recognition Based on ZCPA

Applying difference detection and pruning to emotional speaker recognition

Exploring Spatio-Temporal Representations by Integrating Attention-based Bidirectional-LSTM-RNNs and FCNs for Speech Emotion Recognition

A Study of Zero-Crossings with Peak-Amplitudes in Speech Emotion Classification

ZCPA-based VAP Model in Speech Emotion

Emotional Speech Recognition Based on PAD Emotion Model

Study on Emotion Feature Analysis and Recognition in Speech Signal: an Overview

Speech Emotion Recognition Based on Feature Selection and Extreme Learning Machine Decision Tree

Speech emotion recognition using PCA-based spectral features and HMM

Emotional speaker recognition based on similar neighbor phenomenon

A New Learning Scheme of Emotion Recognition From Speech by Using Mean Fourier Parameters

A Passwords Recognition System Based on ZCPA Feature Parameter

Speech Emotion Recognition using Channel Attention Mechanism

Automatic Emotion Recognition of S

Speech Emotion Recognition Using Emotion Perception Spectral Feature

Study on Method of Emotion Recognition of Speech Based on Feature Parameter Fusion

A Noise-Robust Speech Recognition System Based on ZCPA Features and Support Vector Machine

Research on Robustness of Emotion Recognition under Environmental Noise Conditions

A Study on Emotional Feature Recognition in Speech.

Speech Emotion Recognition Using Acoustic Features

A Robust Feature Extraction Approach Based on an Auditory Model for Classification of Speech and Expressiveness