Search

Automatic structure discovery is desirable in many Markov model applications where a good topology (states and transitions) is not known a priori. CSSR is an established pattern discovery algorithm for stationary and ergodic stochastic symbol sequences that learns a predictively optimal Markov representation consisting of so-called causal states. By means of a novel algebraic criterion, we prove that the causal states of a simple process disturbed by random errors frequently are too complex to be learned fully, making CSSR diverge. In fact, the causal state representation of many hidden Markov models, representing simple but noise-disturbed data, has infinite cardinality. We also report that these problems can be solved by endowing CSSR with the ability to make approximations. The resulting algorithm, robust causal states (RCS), is able to recover the underlying causal structure from data corrupted by random substitutions, as is demonstrated both theoretically and in an experiment. The algorithm has potential applications in areas such as error correction and learning stochastic grammars., QC 20130502

Quantization of the linear predictive coding parameters is an important part in speech coding. Probability density function (PDF)-optimized vector quantization (VQ) has been previously shown to be more efficient than VQ based only on training data. For data with bounded support, some well-defined bounded-support distributions (e.g., the Dirichlet distribution) have been proven to outperform the conventional Gaussian mixture model (GMM), with the same number of free parameters required to describe the model. When exploiting both the boundary and the order properties of the line spectral frequency (LSF) parameters, the distribution of LSF differences (Delta LSF) can be modelled with a Dirichlet mixture model (DMM). We propose a corresponding DMM based VQ. The elements in a Dirichlet vector variable are highly mutually correlated. Motivated by the Dirichlet vector variable's neutrality property, a practical non-linear transformation scheme for the Dirichlet vector variable can be obtained. Similar to the Karhunen-Loeve transform for Gaussian variables, this non-linear transformation decomposes the Dirichlet vector variable into a set of independent beta-distributed variables. Using high rate quantization theory and by the entropy constraint, the optimal inter-and intra-component bit allocation strategies are proposed. In the implementation of scalar quantizers, we use the constrained-resolution coding to approximate the derived constrained-entropy coding. A practical coding scheme for DVQ is designed for the purpose of reducing the quantization error accumulation. The theoretical and practical quantization performance of DVQ is evaluated. Compared to the state-of-the-art GMM-based VQ and recently proposed beta mixture model (BMM) based VQ, DVQ performs better, with even fewer free parameters and lower computational cost., QC 20130815. Updated from submitted to published.

We consider rate optimization in multicast systems that use several multicast trees on a communication network. The network is shared between different applications. For that reason, we model the available bandwidth for multicast as stochastic. For specific network topologies, we show that the multicast rate optimization problem is equivalent to the optimization of scalar quantization. We use results from rate-distortion theory to provide a bound on the achievable performance for the multicast rate optimization problem. A large number of receivers makes the possibility of adaptation to changing network conditions desirable in a practical system. To this end, we derive an analytical solution to the problem that is asymptotically optimal in the number of multicast trees. We derive local optimality conditions, which we use to describe a general class of iterative algorithms that give locally optimal solutions to the problem. Simulation results are provided for the multicast of an i.i.d. Gaussian process, an i.i.d. Laplacian process, and a video source., QC 20120801

The development of objective speech quality measures generally involves fitting a model to subjective rating data. A typical data set comprises ratings generated by listening tests performed in different languages and across different laboratories. These factors as well as others, such as the sex and age of the talker, influence the subjective ratings and result in data heterogeneity. We use a linear hierarchical Bayes (HB) structure to account for heterogeneity. To make the structure effective, we develop a variational Bayesian inference for the linear HB structure that approximates not only the posterior over the model parameters, but also the model evidence. Using the approximate model evidence we are able to study and exploit the heterogeneity inducing factors in the Bayesian framework. The new approach yields a simple linear predictor with state-of-the-art predictive performance. Our experiments show that the new method compares favorably with systems based on more complex predictor structures such as ITU-T recommendation P.563, Bayesian MARS, and Gaussian processes., QC 20120127

We design a K -description scalar quantizer, whose construction is based on a structure of translated scalar lattices and a lattice in K-1 dimensional space. The use of translated lattices provides a performance advantage by exploiting a so-called staggering gain. The use of the K-1 dimensional lattice facilitates analytic insight into the performance and significantly speeds up the computation of the index assignment compared to state-of-the-art methods. Using a common decoding method, the proposed index assignment is proven to be optimal for the K-description case. It is shown that the optimal index assignment is not unique. This is illustrated for the two-description case, where a periodic index assignment is selected from possible optimal assignments and described in detail. The performance of the proposed quantizer accurately matches theoretic analysis over the full range of operational redundancies. Moreover, the quantizer outperforms the state-of-the-art MD scheme as the redundancy among the description increases., QC 20120808

The coding of audio-visual signals is generally based on different paradigms for high and low rates. At high rates the signal is approximated directly and at low rates only signal features are transmitted. The recently introduced distribution preserving quantization (DPQ) paradigm provides a seamless transition between these two regimes. In this paper we present a simplified scheme that preserves the power spectral density (PSD) rather than the probability distribution. In a practical system the PSD must be estimated. We show that both forward adaptive and backward adaptive PSD estimation are possible. Our experimental results confirm that preservation of PSD at finite precision leads to a unified coding paradigm that provides effective coding at both high and low rates. An audio coding application shows the perceptual benefits of PSD preserving quantization., QC 20121119

In this paper, we present a transform-domain Wiener filtering approach for enhancing speech periodicity. The enhancement is performed on the linear prediction residual signal. Two sequential lapped frequency transforms are applied to the residual in a pitch-synchronous manner. The residual signal is effectively represented by two separate sets of transform coefficients that correspond to the periodic and aperiodic components, respectively. A Wiener filter operating on the transform coefficients is developed to restore periodicity and reduce noise. Different filter parameters are designed for the transform coefficients of the periodic and aperiodic components. A template-driven method is used to estimate the filter parameters for the periodic component. For the aperiodic components, the filter parameters are computed based on a local SNR for effective noise reduction. Experimental results confirm that the harmonic structure of the signal can be effectively restored with the proposed approach., QC 20121119

The intelligibility of speech in adverse noise conditions can be improved by modifying the characteristics of the clean speech prior to its presentation. An effective and flexible paradigm is to select the modification by optimizing a measure of objective intelligibility. Here we apply this paradigm at the text level and optimize a measure related to the classification error probability in an automatic speech recognition system. The proposed method was applied to a simple but powerful band-energy modification mechanism under an energy preservation constraint. Subjective evaluation results provide a clear indication of a significant gain in subjective intelligibility. In contrast to existing methods, the proposed approach is not restricted to a particular modification strategy and treats the notion of optimality at a level closer to that of subjective intelligibility. The computational complexity of the method is sufficiently low to enable its use in on-line applications., QC 20130115, Listening Talker (LISTA)

A non-intrusive quality predictor constitutes a mapping from signal features to a (typically one dimensional) representation of the perceived quality. Assuming that the regression model performing the mapping is suited to the data, the performance of the predictor largely depends on how well the parameters of this regression model can be inferred from the training data. In situations where the training data is scarce, model performance is degraded due to over-fitting. The effects of over-fitting can be mitigated by feature selection but the model performance remains low due to the insufficiently representative training data. The objective we pursue is to enhance the performance of a quality predictor by augmenting the feature set with the output of a pre-trained quality predictor. This approach introduces an implicit dependence of the regression model parameters on a larger amount of training data. In view of the increasing usage of speech signals with higher bandwidth, and the dearth of training data for such signals, an augmentation of particular interest is that of a wide-band feature set with a narrow-band quality prediction. Experimental results for additive noise and non-linear distortions encountered in hearing aids, using quality labels from an intrusive quality predictor, illustrate the performance enhancement capabilities of the proposed approach., QC 20121120

In this paper, a novel graph-preserving sparse nonnegative matrix factorization (GSNMF) algorithm is proposed for facial expression recognition. The GSNMF algorithm is derived from the original NMF algorithm by exploiting both sparse and graph-preserving properties. The latter may contain the class information of the samples. Therefore, GSNMF can be conducted as an unsupervised or a supervised dimension reduction method. A sparse representation of the facial images is obtained by minimizing the l(1)-norm of the basis images. Furthermore, according to the graph embedding theory, the neighborhood of the samples is preserved by retaining the graph structure in the mapped space. The GSNMF decomposition transforms the high-dimensional facial expression images into a locality-preserving subspace with sparse representation. To guarantee convergence, we use the projected gradient method to calculate the nonnegative solution of GSNMF. Experiments are conducted on the JAFFE database and the Cohn-Kanade database with unoccluded and partially occluded facial images. The results show that the GSNMF algorithm provides better facial representations and achieves higher recognition rates than nonnegative matrix factorization. Moreover, GSNMF is also more robust to partial occlusions than other tested methods., QC 20110222

QS 20120316

This paper describes a double-ended quality assessment system for speech with a bandwidth of up to 14 kHz (so-called super-wideband speech). The quality assessment system is based on a combination of local and global features, where the local features are dependent on a time alignment procedure and the global features are not. The system is evaluated over a large set of subjectively scored narrowband, wideband and super-wideband speech databases. The system performs similarly to PESQ for narrowband speech and significantly better for wideband speech., QC 20110705

This paper studies the tight rate-distortion bound for L-channel sym-metric multiple-description coding of a vector Gaussian source with twolevels of receivers. Each of the first-level receivers obtains κ (κ < L) ofthe L descriptions. The second-level receiver obtains all L descriptions.We find that when the theory is applied to the scalar Gaussian source, theproduct of a function of the side distortions (corresponding to the first-level receivers) and the central distortion (corresponding to the second-levelreceiver) is asymptotically independent of the redundancy among the de-scriptions. Using this property, we analyze the asymptotic behavior of apractical multiple-description lattice vector quantizer (MDLVQ). Our anal-ysis includes the treatment of the MDLVQ system from a new geometricviewpoint, which results in an expression for the side distortions using thenormalized second moment of a sphere of higher dimensionality than thequantization space. The expression of the distortion product derived fromthe lower bound is then applied as a criterion to assess the performance lossof the considered MDLVQ system. In principle, the efficiency of other prac-tical MD systems can also be evaluated using the derived distortion product.

Using spectral and spectro-temporal auditory models along with perturbation-based analysis, we develop a new framework to optimize a feature vector such that it emulates the behavior of the human auditory system. The optimization is carried out in an offline manner based on the conjecture that the local geometries of the feature vector domain and the perceptual auditory domain should be similar. Using this principle along with a static spectral auditory model, we modify and optimize the static spectral mel frequency cepstral coefficients (MFCCs) without considering any feedback from the speech recognition system. We then extend the work to include spectro-temporal auditory properties into designing a new dynamic spectro-temporal feature vector. Using a spectro-temporal auditory model, we design and optimize the dynamic feature vector to incorporate the behavior of human auditory response across time and frequency. We show that a significant improvement in automatic speech recognition (ASR) performance is obtained for any environmental condition, clean as well as noisy.

The rate region of the two-terminal vector Gaussian CEO problem is studied. A lower bound on the rate region is derived. It is obtained by lower-bounding a weighted sum rate for each supporting hyperplane of the rate region. The bound is in the form of a closed-form expression rather than the form of an optimization problem., QC 20120404

A natural approach to audio coding is to use a rate-distortion optimal design combined with a perceptual model. While this approach is common in transform coding, existing predictive-coding based audio coders are generally not optimal and they benefit from heuristically motivated post-filtering. As delay requirements often force the use of predictive coding, we consider audio coding with a pre- and post-filtered predictive structure that was recently proven to be asymptotically optimal in the rate-distortion sense [1]. We show that this audio coding is efficient in achieving the state-of-the-art performance. We also show that the pre-filter plays a relatively minor role. This leads to an analytic approach for optimizing the post-filter and the predictor at each rate, eliminating the need for manual re-tuning whenever a different rate is called for. In a subjective test, the theoretically optimized post-filter provided a better performance than a conventional post-filter., QC 20120103

Non-intrusive signal quality assessment in general, and its application to speech signal processing, in particular, builds extensively upon statistical regression models. Commonly, the raw preference scores used for fitting these models belong to a categorical scale. Averaging the scores over a number of test subjects results in smooth, close-to-continuous ratings, thus justifying the use of regression as opposed to classification models. A form of marginalization, averaging subjective ratings takes away useful information about the reliability of individual test points. Using a model tailored to the raw data achieves highly competitive performance in terms of conventional performance measures while providing the additional advantage of identifying the usability of individual test points. In this paper, we consider the application of discrete choice models to non-intrusive quality assessment of speech., QC 20130524, Listening Talker (LISTA)

In our previous study of speech periodicity enhancement, the linear prediction residual signal was decomposed into periodic and aperiodic components using two-stage transforms. In the transform domain, the periodic component of the signal is concentrated and represented by a small portion of the coefficients. The respective coefficients were weighted and emphasized to enhance periodicity of the signal against noise. Fixed weights were used for different sets of the coefficients. With the fixed weights, it is observed that unvoiced and voiced-unvoiced transition signals are excessively attenuated and perceptible artificial periodicity are generated in these speech segments. In this study, we propose an adaptive weighting method. For voiced speech, the periodic component is strong and the respective transform coefficients shows a high energy level. In contrast, for unvoiced speech, periodicity is weak and the corresponding coefficients are small. The weights for the coefficients are adaptively adjusted according to the energy level of the periodic component. With the adaptive weights, the periodic component of voiced speech can be effectively emphasized and restored while the aperiodic parts in unvoiced speech are retained., QC 20140828

Using spectral and spectro-temporal auditory models, we develop a computationally simple feature vector based on the design architecture of existing mel frequency cepstral coefficients (MFCCs). Along with the use of an optimized static function to compress a set of filter bank energies, we propose to use a memory-based adaptive compression function to incorporate the behavior of human auditory response across time and frequency. We show that a significant improvement in automatic speech recognition (ASR) performance is obtained for any environmental condition, clean as well as noisy., QC 20110415

A fundamental challenge in the design of objective models for estimation of speech signal quality lies in the shortage of subjectively labelled databases. This problem is particularly relevant when developing quality assessment models for wide-band (16 kHz sampling rate) signals where databases are scarce. We explore the possibility for seamlessly integrating a quality prior in the form of a narrow-band quality estimate into the framework of a non-intrusive wide-band quality assessment algorithm. Experimental results confirm that the proposed approach can be used to improve performance over a baseline wide-band system without a narrow-band prior., QC 20110418

In this work, we study the rate region of the vector Gaussian multipledescription problem with individual and central quadratic distortion con-straints. In particular, we derive an outer bound to the rate region of theL-description problem. The bound is obtained by lower bounding a weightedsum rate for each supporting hyperplane of the rate region. The key ideais to introduce at most L-1 auxiliary random variables and further imposeupon the variables a Markov structure according to the ordering of the de-scription weights. This makes it possible to greatly simplify the derivationof the outer bound. In the scalar Gaussian case, the complete rate regionis fully characterized by showing that the outer bound is tight. In this case,the optimal weighted sum rate for each supporting hyperplane is obtained bysolving a single maximization problem. This contrasts with existing results,which require solving a min-max optimization problem., QC20100830

The rate-distortion performance of conventional resolution-constrained quantization (RCQ) based on the mean-squared error criterion (MSE-RCQ) is generally compromised by the impact of distortion outliers and source mismatch. Not only the mean distortion, but also the number of distortion outliers should be considered in quantizer design. Thus, we propose the use of a design criterion that gives more importance to the tail of the source distribution, which leads to RCQ based on the second moment of distortion (SMD-RCQ). A continuous range of alternatives between MSE-RCQ and SMD-RCQ is also defined and implemented based on the weighted arithmetic-mean measure (WAM-RCQ). It can be used to control the centroid density in the tail of the source distribution. Experimental results with a Gaussian source and line spectral frequencies (LSFs) show that the proposed WAM-RCQ not only produces a similar mean distortion as conventional MSE-RCQ, but has a lower percentage of distortion outliers and a significantly reduced sensitivity to source mismatch., QC 20110329

We propose a probabilistic, non-intrusive method for quality assessment of speech that takes into consideration the bounded character of the preference scores. The quality ratings are modeled as iid Beta random variables, whose mean and precision are parametrized directly in terms of the signal features. Maximum likelihood estimation is used to learn the model parameters in view of a training database. Given a valuation point, the proposed model produces a distribution over the range of allowed quality ratings, which can be used to evaluate the statistics of interest. The model performance, in terms of correlation and root mean square error, compares favorably to the state-of-the-art in the field. Low computational complexity in training and prediction make the model attractive for a wide range of applications. The usage of band-based features in the feature set facilitates extension of the proposed model to input signals with larger bandwidth., QC 20110712

This paper demonstrates the potential of theoretically motivated learning methods in solving the problem of non-intrusive quality estimation for which the state-of-the-art is represented by ITU-T P.563 standard. To construct our estimator, we adopt the speech features from P.563, while we use a different mapping of features to form quality estimates. In contrast to P.563 which assumes distortion-classes to divide the feature space, our approach divides the feature space based on a clustering which is learned from the data using Bayesian inference. Despite using weaker modeling assumptions, we are still able to achieve comparable accuracy on predicting mean-opinion-scores with P.563. Our work suggests Bayesian model-evidence as an alternative metric to correlation-coefficient for determining the necessary number of experts for modeling the data., QC 20110712

Computional learning from multimodal data is often done with matrix factorization techniques such as NMF (Non-negative Matrix Factorization), pLSA (Probabilistic Latent Semantic Analysis) or LDA (Latent Dirichlet Allocation). The different modalities of the input are to this end converted into features that are easily placed in a vectorized format. An inherent weakness of such a data representation is that only a subset of these data features actually aids the learning. In this paper, we first describe a simple NMF-based recognition framework operating on speech and image data. We then propose and demonstrate a novel algorithm that scales the inputs of this framework in order to optimize its recognition performance., QC 20140905

Additive noise generates important losses in automatic speech recognition systems. In this paper, we show that one of the causes contributing to these losses is the fact that conventional recognisers take into consideration feature values that are outliers. The method that we call bounded-distance HMM is a suitable method to avoid that outliers contribute to the recogniser decision. However, this method just deals with outliers, leaving the remaining features unaltered. In contrast, spectral subtraction is able to correct all the features at the expense of introducing some artifacts that, as shown in the paper, cause a larger number of outliers. As a result, we find that bounded-distance HMM and spectral subtraction complement each other well. A comprehensive experimental evaluation was conducted, considering several well-known ASR tasks (of different complexities) and numerous noise types and SNRs. The achieved results show that the suggested combination generally outperforms both the bounded-distance HMM and spectral subtraction individually. Furthermore, the obtained improvements, especially for low and medium SNRs, are larger than the sum of the improvements individually obtained by bounded-distance HMM and spectral subtraction., QC 20100525

A blind two-channel interference reduction algorithm to suppress localised interferers in reverberant environments is presented. The algorithm requires neither knowledge of source positions nor a speech-free noise reference. The goal is to estimate the speech signal as observed at one of the microphones, without any additional filtering effects that are typical in convolutive blind source separation., QC 20100525

A new quantization scheme that preserves the probability distribution of the source signal is presented. The distribution preserving quantization (DPQ) achieves the optimal trade-off between mean square error and bit rate asymptotically. It provides a continuum ranging from rate-distortion optimal signal quantization to parametric coding. The method can be used as a core component for scalable coding. Its efficacy is illustrated by applying the scheme to audio coding., QC 20101213

QS 20120315

We consider using sparse simplifications to denoise probabilistic sequence models for generative tasks such as speech synthesis. Our proposal is to find the least random model that remains close to the original one according to a KL-divergence constraint, a technique we call minimum entropy rate simplification (MERS). This produces a representation-independent framework for trading off simplicity and divergence, similar to rate-distortion theory. Importantly, MERS uses the cleaned model rather than the original one for the underlying probabilities in the KL-divergence, effectively reversing the conventional argument order. This promotes rather than penalizes sparsity, suppressing uncommon outcomes likely to be errors. We write down the MERS equations for Markov chains, and present an iterative solution procedure based on the Blahut-Arimoto algorithm and a bigram matrix Markov chain representation. We apply the procedure to a music-based Markov grammar, and compare the results to a simplistic thresholding scheme., QC 20150708

We propose a new adaptive playout scheme for VoIP. The k- Erlang distribution is introduced to model the packet interarrival time distribution. A cost function is proposed for the next played out packet in the buffer based on modelling packet-arrival times with the k-Erlang distribution. The cost function essentially balances the average buffering delay and the packet-loss rate. The optimal playout length of the packet is determined by minimizing the cost function and realized by either inserting or dropping pitch cycles from the packet. Our real-world data experiments show that our scheme outperforms two reference methods for both low-jitter and highjitter cases., QC 20130528

Periodicity is an important property of speech signals. It plays a critical role in speech communication, especially when strong background noise is present. This paper presents a novel framework of periodicity enhancement for noisy speech. The enhancement operates on the linear prediction error (residual) signal. The residual signal goes through a constant-pitch time warping process and two sequential lapped frequency transforms, by which the periodic component is concentrated in the first modulation band. By emphasizing the respective transform coefficients, periodicity enhancement of noisy residual signal is achieved. The enhanced residual signal and estimated linear prediction filter parameters are used to synthesize the speech output. The effectiveness of the proposed method is confirmed consistently by various objective measures and subjective listening tests. It is observed that the enhanced speech can restore the harmonic structure of the original speech., QC 20140903

This paper presents a packet loss recovery method that uses an incomplete secondary encoding based on scalar quantization as redundancy. The method is redundancy bit rate scalable and allows an adaptation to varying loss scenarios and a varying packeting strategy. The recovery is performed by minimum mean squared error estimation incorporating a statistical model for the quantizers to facilitate real.-time adaptation. A bit allocation algorithm is proposed that extends 'reverse water filling' to the problem of scalar encoding dependent variables for a decoder with a final estimation stage and available side information. We apply the method to the encoding of line-spectral frequencies (LSFs), which are commonly used in speech coding, illustrating the good performance of the method., QC 20100525 32nd IEEE International Conference on Acoustics, Speech and Signal Processing, Honolulu, HI, APR 15-20, 2007

A Bayesian approach to non-intrusive quality assessment of narrow-band speech is presented. The speech features used to assess quality are the sample mean and variance of band-powers evaluated from the temporal envelope in the channels of an auditory filter-bank. Bayesian multivariate adaptive regression splines (BMARS) is used to map features into quality ratings. The proposed combination of features and regression method leads to a high performance quality assessment algorithm that learns efficiently from a small amount of training data and avoids overfitting. Use of the Bayesian approach also allows the derivation of credible intervals on the model predictions, which provide a quantitative measure of model confidence and can be used to identify the need for complementing the training databases., QC 20110221

We consider the optimization of multicast over packet-switched communication networks with a non-zero packet-loss probability. For the system setup consisting of a number of multiple-description coders, we jointly optimize these coders. We propose an analytic solution, asymptotically optimal in the number of multiple-description coders. The analytic solution allows for fast system adaptation to changing network conditions. A locally optimal optimization algorithm that is useful when the number of multicast groups is small is derived. Simulations show that the utilization of the analytic solution incurs a low overhead on the performance when compared to the locally optimal solution, even for a small number of multiple-description coders., QC 20110310

We consider the rate allocation problem for multiple-description quantization of the signal described by an adaptive model with a fixed structure. The source modeling in coding generally results in a two-stage description of the data, where one of the stages describes the model parameters, and the other describes the signal. Such a setup implies the existence of a trade-off between the rate spent on the parameters and the rate spent on the signal. We optimize this trade-off analytically for the multiple-description case using a method inspired by Minimum Description Length principle. We also provide an algorithm for optimizing the rate allocation between the components of the model-based multiple description coder. Finally we experimentally confirm our results. Our method facilitates the rate-adaptive multiple-description coding., QC 20110309

Compressive sensing (CS) has been proposed for signals with sparsity in a linear transform domain. We explore a signal dependent unknown linear transform, namely the impulse response matrix operating on a sparse excitation, as in the linear model of speech production, for recovering compressive sensed speech. Since the linear transform is signal dependent and unknown, unlike the standard CS formulation, a codebook of transfer functions is proposed in a matching pursuit (MP) framework for CS recovery. It is found that MP is efficient and effective to recover CS encoded speech as well as jointly estimate the linear model. Moderate number of CS measurements and low order sparsity estimate will result in MP converge to the same linear transform as direct VQ of the LP vector derived from the original signal. There is also high positive correlation between signal domain approximation and CS measurement domain approximation for a large variety of speech spectra., QC 20110708

In this paper, we present a novel algorithm for representing facial expressions. The algorithm is based on the non-negative matrix factorization (NMF) algorithm, which decomposes the original facial image matrix into two non-negative matrices, namely the coefficient matrix and the basis image matrix. We call the novel algorithm graph-preserving sparse non-negative matrix factorization (GSNMF). GSNMF utilizes both sparse and graph-preserving constraints to achieve a non-negative factorization. The graph-preserving criterion preserves the structure of the original facial images in the embedded sub-space while considering the class information of the facial images. Therefore, GSNMF has more discriminant power than NMF. GSNMF is applied to facial images for the recognition of six basic facial expressions. Our experiments show that GSNMF achieves on average a recognition rate of 93.5% compared to that of discriminant NMF with 91.6%., QC 20110708

Classification on mobile devices is often done in an uninterrupted fashion. This requires algorithms with gentle demands on the computational complexity. The performance of a classifier depends heavily on the set of features used as input variables. Existing feature selection strategies for classification aim at finding a "best" set of features that performs well in terms of classification accuracy, but are not designed to handle constraints on the computational complexity. We demonstrate that an extension of the performance measures used in state-of-the-art feature selection algorithms with a penalty on the feature extraction complexity leads to superior feature sets if the allowed computational complexity is limited. Our solution is independent of a particular classification algorithm., QC 20100525

We present a blind speech watermarking algorithm that embeds the watermark data in the phase of non-voiced speech by replacing the excitation signal of an autoregressive speech signal representation. The watermark signal is embedded in a frequency subband, which facilitates robustness against bandpass filtering channels. We derive several sets of pulse shapes that prevent inter-symbol interference and that allow the passband watermark signal to be created by simple filtering. A marker-based synchronization scheme robustly detects the location of the embedded watermark data without the occurrence of insertions or deletions. In light of a potential application to analog aeronautical voice radio communication, we present experimental results for embedding a watermark in narrowband speech at a bit-rate of 450 bit/s. The recursive least-squares (RLS) equalization-based watermark detector not only compensates for the vocal tract filtering, but also recovers the watermark data in the presence of nonlinear phase and bandpass filtering, amplitude modulation, and additive white Gaussian noise (AWGN), making the watermarking scheme highly robust., QC 20110415

We address optimal model estimation for model-based vector quantization for both the constrained resolution (CR) and constrained entropy (CE) cases. To this purpose we derive under high-rate (HR) theory assumptions the rate-distortion (RD) relations for these two quantization scenarios assuming a Gaussian model. Based on the RD relations we show that the maximum likelihood (ML) criterion leads to optimal performance for CE quantization, but not for CR quantization. We introduce a new model estimation criterion for CR quantization that is optimal (under HR theory assumptions) in terms of the RD relation. Our experiments confirm that the proposed criterion for model identification outperforms the ML criterion for a range of conditions., QC 20110708

This paper studies the tight rate-distortion bound for K-channel symmetric multiple-description coding for a memoryless Gaussian source. We find that the product of a function of the individual side distortions (for single received descriptions) and the central distortion (for K received descriptions) is asymptotically independent of the redundancy among the descriptions. Using this property, we analyze the asymptotic behaviors of two different practical multiple-description lattice vector quantizers (MDLVQ). Our analysis includes the treatment of a MDLVQ system from a new geometric viewpoint, which results in an expression for the side distortions using the normalized second moment of a sphere of higher dimensionality than the quantization space. The expression of the distortion product derived from the lower bound is then applied as a criterion to assess the performance losses of the considered MDLVQ systems., QC 20110318

In this paper camera pose control for optimizing multiview distributed video coding is considered. The scenario considered is that multiple agents with monocular cameras observe a common scene in a three dimensional world. To get a good video reconstruction under a transmission rate constraint, the camera closest to the center of the camera array is chosen as the reference camera. The poses of all other cameras are controlled and designed such that their images are maximally similar under a constraint on their separation. Based on the rigid motion allowed for the cameras, two cases are considered. For the case where the rigid motion only involves translation, translation control is designed. For the case where it involves both translation and rotation, both controls are constructed. Some simulated results are given to show the efficiency of the designed controllers., QC 20110711

Traditionally, sound codecs have been developed with a particular application in mind, their performance being optimized for specific types of input signals, such as speech or audio (music), and application constraints, such as low bit rate, high quality, or low delay. There is, however, an increasing need for more generic sound codecs, created by the emergence of heterogeneous networks and the convergence of communication and entertainment devices. To obtain such versatility, this study employs hybrid sound coding based on operational rate-distortion (RD) optimization principles. Applying this concept, a prototype coder has been implemented with emphasis on (dynamic) adaptation to the input and to application constraints. With this prototype, listening tests have been performed for different application scenarios. The results demonstrate the versatility of the concept while keeping competitive sound quality compared to dedicated state-of-the-art codecs., QC 20100525

All-pole spectral envelope estimates based on linear prediction (LP) for speech signals often exhibit unnaturally sharp peaks, especially for high-pitch speakers. In this paper, regularization is used to penalize rapid changes in the spectral envelope, which improves the spectral envelope estimate. Based on extensive experimental evidence, we conclude that regularized linear prediction outperforms bandwidth-expanded linear prediction. The regularization approach gives lower spectral distortion on average, and fewer outliers, while maintaining a very low computational complexity., QC 20100525

We consider adaptive two-channel multiple-description coding. We provide an analytical method for designing a resolution-constrained symmetrical multiple-description coder that uses an index assignment matrix. We use existing index assignment algorithms that are known for their good properties within our adaptive multiple-description coding architecture. These existing index assignment algorithms are parameterized and the coefficient of quantization of the side coders is described by a rational function. This leads to an analytical solution for the design problem, facilitating real-time adaptation based on information provided by feedback channels or rate conditions imposed by network management. Our experimental results show that the practical performance closely approximates theoretically obtained optimal behavior., QC 20110708

We consider the problem of multicast rate optimization for a given number of multicast groups. The efficient solution of this problem is particularly relevant for video multicast. Motivated by practical considerations, such as the need for adaptivity to changing network conditions, we develop an analytical solution that minimizes the expected distortion of a receiver. Unlike previous work, we recognize variability in existing networks and model the receiver reception capacities by a continuous stochastic variable. Our analytical solution is optimal under the assumption of a large number of multicast groups. We investigate the distortion overhead when this assumption is not satisfied and find that it is low. We compare our results to the state of the art, iterative optimization for a realization of receiver reception capacities, and find that our analytical solution performs better for any number of multicast groups. copyright by EURASIP., QC 20141017

Mutual information is commonly used in speech processing in the context of statistical mapping. Examples are the optimization of speech or speaker recognition algorithms, the computation of performance bounds on such algorithms, and bandwidth extension of narrow-band speech signals. It is generally ignored that speech-signal derived data usually have an intrinsic dimensionality that is lower than the dimensionality of the observation vectors (the dimensionality of the embedding space). In this paper, we show that such reduced dimensionality can affect the accuracy of the mutual information estimate significantly. We introduce a new method that removes the effects of singular probability density functions. The method does not require prior knowledge of the intrinsic dimensionality of the data. It is shown that the method is appropriate for speech-derived data., Book Group Author(s): ISCA