Many speech emotion recognition systems have been designed using different features and classification methods. Still, there is a lack of knowledge and reasoning regarding the underlying speech characteristics and processing, i.e., how basic characteristics, methods, and settings affect the accuracy, to what extent, etc. This study is to extend physical perspective on speech emotion recognition by analyzing basic speech characteristics and modeling methods, e.g., time characteristics (segmentation, window types, and classification regions—lengths and overlaps), frequency ranges, frequency scales, processing of whole speech (spectrograms), vocal tract (filter banks, linear prediction coefficient (LPC) modeling), and excitation (inverse LPC filtering) signals, magnitude and phase manipulations, cepstral features, etc. In the evaluation phase the state-of-the-art classification method and rigorous statistical tests were applied, namely N-fold cross validation, paired t-test, rank, and Pearson correlations. The results revealed several settings in a 75% accuracy range (seven emotions). The most successful methods were based on vocal tract features using psychoacoustic filter banks covering the 0–8 kHz frequency range. Well scoring are also spectrograms carrying vocal tract and excitation information. It was found that even basic processing like pre-emphasis, segmentation, magnitude modifications, etc., can dramatically affect the results. Most findings are robust by exhibiting strong correlations across tested databases.

On the Speech Properties and Feature Extraction Methods in Speech Emotion Recognition.

Speech Emotion Recognition: A Comprehensive Survey

Emotion recognition from speech signal has become more and more important in advanced human-machine applications. The detailed description of emotions and their detection play an important role in the psychiatric studies but also in other fields of medicine such as anamnesis, clinical studies or lie detection. In this paper some experiments using multilingual emotional databases are presented. For the features extracted from the speech material, the LPC (Linear predictive coding), LPCC (Linear Predictive Cepstral Coefficients) and MFCC (Mel Frequency Cepstral Coefficients) coefficients are employed. The Weka tool was used for the classification task, selecting the k-NN (k-nearest neighbors) and SVM (Support Vector Machine) classifiers. The results for the selected features vectors show that the emotion recognition rate is satisfactory when multilingual speech material is used for training and testing. When the training is made using emotional materials for a language and testing with materials in other language the results are poor. Therefore, this shows that the features extracted from speech display a closed dependency with the spoken language.

Corina Albu

Papers

Emotion Recognition from Speech Signal in Multilingual Experiments