Factor Analyzed Subspace Modeling and Selection

Chien, Jen‐Tzung; Ting, Chuan-Wei

doi:10.1109/tasl.2007.910790

Cited by 11 publications

(7 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These hyperparameters are seen as the ARD parameters or the regularization parameters. The group representation of speech features can be seen as a kind of subspace approach [30]. The common bases span the principal subspace for representing the principal information within a regression class while the state-dependent individual bases span the minor subspace for catching the residual information due to an individual state.…”

Section: B Comparison With Other Modelsmentioning

confidence: 99%

“…MAP estimate is calculated by using observation and current estimates , , , , and . Similarly, MAP estimate of the th sensing weight for individual basis at frame is derived by collecting all terms related to and minimizing the resulting -regularized objective function to yield (30) for . MAP estimate is calculated by using and current estimates , , , , and .…”

Section: Estimation Of Sensing Weights and Hessian Matrixmentioning

confidence: 99%

See 1 more Smart Citation

Laplace Group Sensing for Acoustic Models

Chien

2015

IEEE/ACM Trans. Audio Speech Lang. Process.

Self Cite

View full text Add to dashboard Cite

This paper presents the group sparse learning for acoustic models where a sequence of acoustic features is driven by Markov chain and each feature vector is represented by groups of basis vectors. The group of common bases represents the features across Markov states within a regression class. The group of individual basis compensates the intra-state residual information. Laplace distribution is used as the sparse prior of sensing weights for group basis representation. Laplace parameter serves as regularization parameter or automatic relevance determination which controls the selection of relevant bases for acoustic modeling. The groups of regularization parameters and basis vectors are estimated from training data by maximizing the marginal likelihood over sensing weights which is implemented by Laplace approximation using the Hessian matrix and the maximum a posteriori parameters. Model uncertainty is compensated through full Bayesian treatment. The connection of Laplace group sensing to lasso regularization is illustrated. Experiments on noisy speech recognition show the robustness of group sparse acoustic models in presence of different noise types and SNRs.

show abstract

Section: B Comparison With Other Modelsmentioning

confidence: 99%

Section: Estimation Of Sensing Weights and Hessian Matrixmentioning

confidence: 99%

Laplace Group Sensing for Acoustic Models

Chien

2015

IEEE/ACM Trans. Audio Speech Lang. Process.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Such a signal recovery problem could be interpreted from a perspective of subspace approach. Namely, an observed signal is demixed into one signal from principal subspace spanned by common bases and the other signal from minor subspace spanned by individual bases [31]. Moreover, the sparseness constraint is imposed on two groups of reconstruction weights S…”

Section: Model Constructionmentioning

confidence: 99%

Bayesian group sparse learning for music source separation

Chien

Hsieh

2013

J AUDIO SPEECH MUSIC PROC.

Self Cite

View full text Add to dashboard Cite

Nonnegative matrix factorization (NMF) is developed for parts-based representation of nonnegative signals with the sparseness constraint. The signals are adequately represented by a set of basis vectors and the corresponding weight parameters. NMF has been successfully applied for blind source separation and many other signal processing systems. Typically, controlling the degree of sparseness and characterizing the uncertainty of model parameters are two critical issues for model regularization using NMF. This paper presents the Bayesian group sparse learning for NMF and applies it for single-channel music source separation. This method reconstructs the rhythmic or repetitive signal from a common subspace spanned by the shared bases for the whole signal and simultaneously decodes the harmonic or residual signal from an individual subspace consisting of separate bases for different signal segments. A Laplacian scale mixture distribution is introduced for sparse coding given a sparseness control parameter. The relevance of basis vectors for reconstructing two groups of music signals is automatically determined. A Markov chain Monte Carlo procedure is presented to infer two sets of model parameters and hyperparameters through a sampling procedure based on the conditional posterior distributions. Experiments on separating single-channel audio signals into rhythmic and harmonic source signals show that the proposed method outperforms baseline NMF, Bayesian NMF, and other group-based NMF in terms of signal-to-interference ratio.

show abstract

“…2 The change point detection between speech and non-speech is estimated by the phoneme recognition using the trained HMM/GMM model. 3 The variance of the noise is updated when the non-speech detected, a priori and a posterior of each frame are then calculated using the Equation (15) and (16). 4 The estimationÂ k (n) is calculated using the Equation (14).…”

Section: A Recursive Phoneme Recognition and Speech Enhancement Methomentioning

confidence: 99%

“…Voice activity detection (VAD), which is a scheme to detect the presence of speech in the observed signals automatically, plays an important role in speech signal processing [1][2][3][4]. It is because that high accurate VAD can reduce bandwidth usage and network traffic in voice over IP (VoIP), and can improve the performance of speech recognition in noisy systems.…”

Section: Introductionmentioning

confidence: 99%

A novel voice activity detection based on phoneme recognition using statistical model

Bao

Zhu

2012

J AUDIO SPEECH MUSIC PROC.

View full text Add to dashboard Cite

In this article, a novel voice activity detection (VAD) approach based on phoneme recognition using Gaussian Mixture Model based Hidden Markov Model (HMM/GMM) is proposed. Some sophisticated speech features such as high order statistics (HOS), harmonic structure information and Mel-frequency cepstral coefficients (MFCCs) are employed to represent each speech/non-speech segment. The main idea of this new method is regarding the non-speech as a new phoneme corresponding to the conventional phonemes in mandarin, and all of them are then trained under maximum likelihood principle with Baum-Welch algorithm using GMM/HMM model. The Viterbi decoding algorithm is finally used for searching the maximum likelihood of the observed signals. The proposed method shows a higher speech/non-speech detection accuracy over a wide range of SNR regimes compared with some existing VAD methods. We also propose a different method to demonstrate that the conventional speech enhancement method only with accurate VAD is not effective enough for automatic speech recognition (ASR) at low SNR regimes.

show abstract

Factor Analyzed Subspace Modeling and Selection

Cited by 11 publications

References 31 publications

Laplace Group Sensing for Acoustic Models

Laplace Group Sensing for Acoustic Models

Bayesian group sparse learning for music source separation

A novel voice activity detection based on phoneme recognition using statistical model

Contact Info

Product

Resources

About