A DCT based nonlinear predictive coding for feature extraction in speech recognition systems

Azar, Mahmood Yousefi; Razzazi, Farbod

doi:10.1109/cimsa.2008.4595825

Cited by 6 publications

(3 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…DCT may be extracted directly or through a neural network [18]. Our primary results on using DCT have been published in [19]. In this approach, DCT is calculated directly.…”

Section: Dct-npcmentioning

confidence: 99%

A neural predictive coding feature extraction scheme in DCT domain for phoneme recognition

Azar

Razzazi

2010

Neural Comput & Applic

Self Cite

View full text Add to dashboard Cite

Nonlinear feature extraction of speech signals has been the main concern of many researches in recent years. In this paper, feature extraction of phonemes using NPC (neural predictive coding) model is generalized to a combination of time and DCT domains. Two main ideas were proposed and evaluated in this paper. First, a framewise DCT-based NPC feature extractor is proposed to overcome the computational complexity deficiency of the system. The basis of this approach is the application of a DCT pre-feature extractor to remove unwanted additional data. In this approach, the extracted features are the output of the hidden layer. It is shown that the use of a preprocessing stage can improve both computational complexity efficiency and accuracy issues. At the second approach, we proposed a complementary role for DCT domain features in classic NPC modeling. This approach uses the signal residual of the predicted signal in the DCT domain. The experiments were conducted on voiced plosive phonemes of TIMIT database. Simulations showed that the performance of the combined method is good at the plosive phonemes. The achieved accuracy that was resulted from the proposed method was 70.3% recognition rate on / b/d/g/ phonemes, which is higher than the results of traditional NPC approaches.

show abstract

“…DCT may be extracted directly or through a neural network [18]. Our primary results on using DCT have been published in [19]. In this approach, DCT is calculated directly.…”

Section: Dct-npcmentioning

confidence: 99%

A neural predictive coding feature extraction scheme in DCT domain for phoneme recognition

Azar

Razzazi

2010

Neural Comput & Applic

Self Cite

View full text Add to dashboard Cite

show abstract

“…3) Finally, the matrices of mean (5) and variances (6) are generated. The parameters of and are used to produce Gaussians matrices which will be used as fundamental information for implementation of the fuzzy recognition system.…”

Section: A Two-dimensional Time Matrix Dct Codingmentioning

confidence: 99%

A novel intelligent system for speech recognition

Silva¹,

Serra²

2014

2014 International Joint Conference on Neural Networks (IJCNN)

View full text Add to dashboard Cite

The concept of fuzzy sets and fuzzy logic is widely used to propose several methods applied to modeling, classification and pattern recognition problem. This paper proposes an Intelligent Methodology for Speech Recognition (IMSR). In addition to pre-processing, with mel-cepstral coefficients, the Discrete Cosine Transform (DCT) is used to generate a two-dimensional time matrix for each pattern to be recognized. A genetic algorithm is used to optimize a Mamdani fuzzy inference system in order to obtain the best model with minimum number of parameters for final recognition. Experimental results for speech recognition applied to brazilian language show the efficiency of the proposed methodology compared to methodologies widely used and cited in the literature.

show abstract

“…where m f cc k are the mel-cepstral coefficients, and k, 1 ≤ k ≤ K, is the k-th (line) component of t-th frame of the matrix and n, 1 ≤ n ≤ N (column) is the order of DCT. Thus, the two-dimensional time matrix (Azar and Razzazi, 2008), where the interesting loworder coefficients k and n that encode the long-term variations of the spectral envelope of the speech signal is obtained (Fissore et al, 1997). Thus, there is a two-dimensional time matrix C k (n, T ) for each input speech signal.…”

Section: Two-dimensional Time Matrix Dct Codingmentioning

confidence: 99%

GFIS: Genetic Fuzzy Inference System for Speech Recognition

Silva¹,

Serra²

2012

Proceedings of the 9th International Conference on Informatics in Control, Automation and Robotics

View full text Add to dashboard Cite

The concept of fuzzy sets and fuzzy logic is widely used to propose of several methods applied to systems modeling, classification and pattern recognition problem. This paper proposes a genetic-fuzzy recognition system for speech recognition. In addition to pre-processing, with mel-cepstral coefficients, the Discrete Cosine Transform (DCT) is used to generate a two-dimensional time matrix for each pattern to be recognized. A genetic algorithms is used to optimize a Mamdani fuzzy inference system in order to obtain the best model for final recognition. The speech recognition system used in this paper was named Genetic Fuzzy Inference System for Speech Recognition (GFIS). Experimental results for speech recognition applied to brazilian language show the efficiency of the proposed methodology compared to methodologies widely used and cited in the literature. 536 Luis Santos Silva W. and Serra G..

show abstract

A DCT based nonlinear predictive coding for feature extraction in speech recognition systems

Cited by 6 publications

References 11 publications

A neural predictive coding feature extraction scheme in DCT domain for phoneme recognition

A neural predictive coding feature extraction scheme in DCT domain for phoneme recognition

A novel intelligent system for speech recognition

GFIS: Genetic Fuzzy Inference System for Speech Recognition

Contact Info

Product

Resources

About