SVM-MLP-PNN Classifiers on Speech Emotion Recognition Field - A Comparative Study

Iliou, Theodoros; Anagnostopoulos, Christos‐Nikolaos

doi:10.1109/icdt.2010.8

Cited by 22 publications

(17 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Some emotions are overlapped and confusing. Few basic emotions as anger, happiness, sadness, boredom, surprise and fear has been found in literatures more often [1][2][3][4][5]. However, emotions as joy, disgust, anxiety and similar nature, emotions out shadow these basic emotions.…”

Section: Introductionmentioning

confidence: 90%

“…Statistics as mean, median, standard deviation, maximum, minimum, range, etc., of these features have been attempted in this field. Features as the fundamental frequency, formants, amplitude, energy, speech rate, Linear Prediction Coefficient (LPC), Mel-Frequency Cepstral Coefficient (MFCC), Perceptual Linear Prediction (PLP) and other spectral variants have been found in literature [1][2][3][4][5][6]. Selection of suitable feature is one of the basic design criteria for emotion detection system.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Reduced feature extraction for emotional speech recognition

Palo

Mohanty

2015

2015 Annual IEEE India Conference (INDICON)

View full text Add to dashboard Cite

There has been a considerable use of acoustic features for speaker identification and recognition. Few of these features have also been used by researchers to recognize emotions in speech effectively. Here an attempt is made to characterize human speech emotions with acoustic features as speech rate, formant frequencies, amplitude and energy initially. Further, a reduced acoustic feature set based on difference between consecutive feature values has been proposed. Multilayer Perceptron (MLP) classifier has been put to test for simulation with both baseline and reduced feature sets. The proposed feature sets outperformed baseline features in terms of classification accuracy and Mean Square Error (MSE) as manifested in the results.

show abstract

Section: Introductionmentioning

confidence: 90%

Section: Introductionmentioning

confidence: 99%

Reduced feature extraction for emotional speech recognition

Palo

Mohanty

2015

2015 Annual IEEE India Conference (INDICON)

View full text Add to dashboard Cite

show abstract

“…There are only a few studies which describe these two approaches in one study (e.g., Cichosz and Slot, 2007;Iliou and Anagnostopoulos, 2010); however, precise comparisons detailing the amount of speaker data used in training were not described.…”

Section: Aim Of This Studymentioning

confidence: 99%

“…The confusion matrices also showed similar tendencies. Iliou and Anagnostopoulos (2010) claim that they attained 84% using a speaker dependent configuration, SVMs and seven emotions. However, various details of this study remain unclear, such as how it was possible to achieve speaker dependent conditions with only 45 recordings of "disgust" for all 10 speakers and using 10-fold cross-validation.…”

Section: 1mentioning

confidence: 99%

“…The output of a neuron is usually activated by a non-linear activation function (for example, sigmoid) based on a weighted sum of the inputs of the neuron. ANNs can have various architectures, of which the feed-forward one is mostly met in emotion recognition (e.g., Iliou and Anagnostopoulos, 2010;Yacoub et al, 2003). ANNs are usually trained in supervised mode, using the back-propagation algorithm.…”

Section: 4mentioning

confidence: 99%

See 1 more Smart Citation

Comparison of speaker dependent and speaker independent emotion recognition

Rybka

Janicki

2013

International Journal of Applied Mathematics and Computer Science

View full text Add to dashboard Cite

This paper describes a study of emotion recognition based on speech analysis. The introduction to the theory contains a review of emotion inventories used in various studies of emotion recognition as well as the speech corpora applied, methods of speech parametrization, and the most commonly employed classification algorithms. In the current study the EMO-DB speech corpus and three selected classifiers, the k-Nearest Neighbor (k-NN), the Artificial Neural Network (ANN) and Support Vector Machines (SVMs), were used in experiments. SVMs turned out to provide the best classification accuracy of 75.44% in the speaker dependent mode, that is, when speech samples from the same speaker were included in the training corpus. Various speaker dependent and speaker independent configurations were analyzed and compared. Emotion recognition in speaker dependent conditions usually yielded higher accuracy results than a similar but speaker independent configuration. The improvement was especially well observed if the base recognition ratio of a given speaker was low. Happiness and anger, as well as boredom and neutrality, proved to be the pairs of emotions most often confused.

show abstract