Intelligent system of speech recognition using Neural Networks based on DCT parametric models of low order

Rocha, Priscila Lima; Silva, Washington Luis Santos

doi:10.1109/ijcnn.2016.7727280

Cited by 5 publications

(3 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are also the cases of applications for embedded projects with restricted vocabulary size, in which there is no need for many examples to train them and can be used on a digital signal processor (DSP) hardware. Thus, the conventional neural networks are presented as a good tool, as shown in the works of Rocha and Silva (2016). However, it was observed the limitation of these neural networks with the expansion of the number of classes to be recognized.…”

Section: Motivation and Justificationmentioning

confidence: 99%

Hierarchical Expert Neural Network System for Speech Recognition

Rocha

Silva

Barros

2019

J Control Autom Electr Syst

View full text Add to dashboard Cite

This work proposes a hierarchical architecture composed of a expert neural network set based on the ensemble method with dynamic selection of classifiers for application in speech recognition systems. Therefore, 30 commands in the Brazilian Portuguese language were coded by a two-dimensional time matrix, resulting from the application of the discrete cosine transformation in the mel-cepstral coefficients. These patterns were modified by means of a nonlinear transformation to a high-dimensionality space through a set of Gaussian radial basis functions (GRBFs) parameterized with the centroid and covariance characteristics of the classes. The classification was made through the dynamic classifier selection approach, in which multilayer perceptron and learning vector quantization configurations were analyzed to constitute the multiple classifiers specialized in the subdivisions made in the total of classes to be recognized. Then, given a new test pattern, the GRBF that presents the highest value of the receptive field in relation to the input feature vector indicates the class to which the pattern is nearer, thus directing to the expert neural network that provides the final result of classification based on the local accuracy.

show abstract

Section: Motivation and Justificationmentioning

confidence: 99%

Hierarchical Expert Neural Network System for Speech Recognition

Rocha

Silva

Barros

2019

J Control Autom Electr Syst

View full text Add to dashboard Cite

show abstract

“…For the structure of the LVQ neural network, it was necessary to define the η learning rate and the n number of neurons of the competitive layer. The defined values in η set are often used in the specialized literature [17,18,26] and the n set was specified considering that the number of neurons in hidden layer should be greater than the number of inputs and greater than the number of neural network outputs. Because the vectors C N Jm , where N = {4, 9, 16} are mapped into a 30-dimensional space, the input of 15 LVQ experts is a set with 30 source nodes.…”

Section: Lvq Expertsmentioning

confidence: 99%

“…The robustness provided in the classification task is a result of the inherent adaptive characteristic of neural networks, allowing them to be able to learn complex patterns and trends present in the set of data available for identification, changing rapidly to modifications in the environment in which is inserted [13][14][15]. The neural networks have several configurations for solution of the most problems and among such configurations with the best results in solving pattern classification problems are multilayer perceptron (MLP) and the learning vector quantization (LVQ) [16,17].…”

Section: Introductionmentioning

confidence: 99%

Neural Network Configurations Analysis for Multilevel Speech Pattern Recognition System with Mixture of Experts

Silva¹,

Rocha²,

Filho³

2018

Intelligent System

Self Cite

View full text Add to dashboard Cite

This chapter proposes to analyze two configurations of neural networks to compose the expert set in the development of a multilevel speech signal pattern recognition system of 30 commands in the Brazilian Portuguese language. Then, multilayer perceptron (MLP) and learning vector quantization (LVQ) networks have their performances verified during the training, validation and test stages in the speech signal recognition, whose patterns are given by two-dimensional time matrices, result from mel-cepstral coefficients coding by the discrete cosine transform (DCT). In order to avoid the pattern separability problem, the patterns are modified by a nonlinear transformation to a high-dimensional space through a suitable set of Gaussian radial base functions (GRBF). The performance of MLP and LVQ experts is improved and configurations are trained with few examples of each modified pattern. Several combinations were performed for the neural network topologies and algorithms previously established to determine the network structures with the best hit and generalization results.

show abstract