Improved Voice Activity Detection based on support vector machine with high separable speech feature vectors

Zou, Yuexian; Zheng, Weiqiao; Shi, Wei; Liu, Hong

doi:10.1109/icdsp.2014.6900767

Cited by 16 publications

(11 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…A detecção de atividade vocal (VAD -Voice Activity Detection)é um assunto que tem recebido a atenção da comunidade científica por muitos anos [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15]. Em sistemas de telefonia, detectores de atividade vocal permitem uma significativa redução na largura de banda utilizada para comunicações de voz.…”

Section: Introductionunclassified

“…Nesse trabalho, uma Máquina de Vetor de Suporte (SVM -Support Vector Machine) recebe as medidas usadas pelo VAD G.729-B como entradas, obtendo resultados superiores aos do G.729-B convencional. Em outra abordagem, o mecanismo de aprendizagemé alimentado com medidas que buscam representar o som propriamente dito, como seria o caso dos coeficientes mel-cepstrais (MFCCs -Mel-frequency cepstral coefficients) [14]. Assim, o classificador torna-se responsável não apenas pela decisão final do detector, mas também pela descoberta das características importantes para discriminar a presença ou ausência de voz.…”

Section: Introductionunclassified

See 1 more Smart Citation

Detecção de Atividade Vocal com o uso de Máquinas de Boltzmann Restritas Discriminativas

Rogério¹,

Magno²

2016

Anais De XXXIV Simpósio Brasileiro De Telecomunicações

View full text Add to dashboard Cite

Resumo-A detecção de atividade vocal (VAD-Voice Activity Detection) tem apreciável impacto no desempenho de aplicações tecnológicas atuais, tais como comunicações sem fio e reconhecimento de fala. Neste trabalho, aborda-se a tarefa de VAD por meio da aprendizagem de máquina usando-se uma estrutura denominada Máquina de Boltzmann Restrita Discriminativa (DRBM-Discriminative Restricted Boltzmann Machine). Com o emprego de vetores de características baseados tanto em coeficientes mel-cepstrais quanto em energias em bandas do espectro, chega-se a desempenhos levemente superiores aos do VAD denominado Long-Term Spectral Divergence (LTSD), usado frequentemente como base de comparação para outros detectores. Os resultados indicam ainda que a DRBMé capaz de lidar com vetores de características fortemente correlacionados.

show abstract

Section: Introductionunclassified

Detecção de Atividade Vocal com o uso de Máquinas de Boltzmann Restritas Discriminativas

Rogério¹,

Magno²

2016

Anais De XXXIV Simpósio Brasileiro De Telecomunicações

View full text Add to dashboard Cite

show abstract

“…Zhang et al(2013) [26] introduced a multiple-feature fusion method using deep-belief networks (DBNs) for VAD by exploiting the deep model to combine multiple features nonlinearly. Zou et al(2014) [27] exploited the MFCC features and SVM classifier for VAD and Tu et al(2014) [28] proposed VAD methods using discriminative acoustic features from computational auditory scene analysis.…”

Section: Introductionmentioning

confidence: 99%

An evolutionary approach for segmentation of noisy speech signals for efficient voice activity detection

Sattar

Rudzicz

Pwint³

2015

AIR

View full text Add to dashboard Cite

This paper presents a new approach to automatically segmenting speech signals in noisy environments. Segmentation of speech signals is formulated as an optimization problem and the boundaries of the speech segments are detected using a genetic algorithm (GA). The number of segments present in a signal is initially estimated from the reconstructed sequence of the original signal using the minimal number of Walsh basis functions. A multi-population GA is then employed to determine the locations of segment boundaries. The segmentation results are improved through the generations by introducing a new evaluation function which is based on the sample entropy and a heterogeneity measure. Experimental results show that the proposed approach can accurately detect the noisy speech segments as well as noise-only segments under various noisy conditions.

show abstract

“…This is the case of the scheme proposed in [13], where a support vector machine (SVM) was fed with measurements of G.729-B to improve the performance of the conventional VAD. In some other approaches, the learning mechanism is fed with measurements that represent the sound, as mel-frequency cepstral coefficients [14]. In this case, besides taking the final decision, the classifier is also responsible for discovering the features that should be used to decide between presence or absence of speech.…”

Section: Introductionmentioning

confidence: 99%

Voice activity detection using discriminative restricted Boltzmann machines

Borin

Silva

2017

2017 25th European Signal Processing Conference (EUSIPCO)

View full text Add to dashboard Cite

Abstract-Voice Activity Detection (VAD) plays an important role in current technological applications, such as wireless communications and speech recognition. In this paper, we address the VAD task through machine learning by using a discriminative restricted Boltzmann machine (DRBM). We extend the conventional DRBM to deal with continuous-valued data and employ feature vectors based either on mel-frequency cepstral coefficients or on filter-bank energies. The resulting detector slightly outperforms the VAD often used as benchmark for detector comparison. Results also indicate that DRBM is able to deal with strongly correlated feature vectors.

show abstract

Improved Voice Activity Detection based on support vector machine with high separable speech feature vectors

Cited by 16 publications

References 19 publications

Detecção de Atividade Vocal com o uso de Máquinas de Boltzmann Restritas Discriminativas

Detecção de Atividade Vocal com o uso de Máquinas de Boltzmann Restritas Discriminativas

An evolutionary approach for segmentation of noisy speech signals for efficient voice activity detection

Voice activity detection using discriminative restricted Boltzmann machines

Contact Info

Product

Resources

About