Analysis-synthesis based speech enhancement with improved spectrum envelope estimation by tracking speech dynamics

This paper presents a method for interpolation of lost speech segments. The interpolation method can be used for packet loss concealment in voice communication over mobile phones, for voice over IP or for restoration of lost segments in speech recordings. The interpolation method employs a combination of a linear prediction (LP) model of the spectral envelope and a harmonic noise model (HNM) of the excitation of speech. The speech interpolation problem is transformed to the modeling and interpolation of the trajectories of LP parameters and the amplitude, phase and harmonicity of HNM tracks of speech excitation. In particular, the interpolation of harmonicity results in a smooth transition from voiced to unvoiced speech and vice versa. Crucially, the proposed interpolation method does not suffer from the consequences of zero-excitation of conventional autoregressive (AR) interpolation. Different combinations of linear and autoregressive interpolation methods are evaluated for the estimation of the time-varying parameters of LP-HNM tracks. Furthermore, a post-processing codebook mapping, employed to enhance the interpolation of the spectral envelope of speech, results in improved output quality for longer length speech gaps. For different packet loss rates and patterns of distributions of missing speech gaps, the proposed interpolation methods are evaluated and compared with popular AR-based interpolation methods and the speech packet recovery method specified in the ITU G.711 standard, as a reference. The evaluation results show that the proposed methods substantially improve the restoration of formants and harmonic tracks and consistently results in significant performance gain and improved perceptual quality of speech.

This paper proposes an algorithm that adopts the harmonic regeneration as post-processing to improve the performance of speech enhancement using traditional Short Time Spectral Amplitude (STSA). The proposed algorithm aims to alleviate the distortion of the high harmonics of enhanced speech via the traditional STSA, and consequently improves the speech quality. We first detect the pitch, or fundamental frequency, of the enhanced speech via the traditional STSA, and then, divide the whole spectrum into multiple sub-bands which center on each harmonic. After that, a series of specially designed windows centered on each harmonic are applied to all the sub-bands, in order to redistribute the energy in the sub-bands. The results of experiment demonstrate that the method has both theoretical and practical basis.

Undirected graphical models such as Markov random fields or Boltzmann machines prove useful in many signal processing and machine learning tasks. However, parameter estimation in these models is difficult due to the intractable normalising constant in their probability density functions. One powerful technique for parameter estimation in such models is score matching. This technique makes use of an objective function which is independent of the normalising constant and constitutes locally consistent estimators for the parameters of such models. However, score matching is only applicable to fully-observed models. In this paper, we extend the applicability of score matching to models with latent variables. Our estimators are unbiased, based on Monte Carlo integration. Unbiased gradient estimators open the way to optimisation through stochastic approximation. We demonstrate the performance of our methodology on two synthetic problems.

Subspace-based direction of arrival estimation in colored ambient noise environments

In this paper, we discuss the amount of musical noise generation for blind speech extraction using minimum mean-square error short-time spectral amplitude (MMSE-STSA) estimator. To achieve high quality speech enhancement, we have previously proposed blind spatial subtraction array (BSSA). However, BSSA always suffers from artificial distortion, so-called musical noise, owing to nonlinear signal processing. Therefore, we propose the improved BSSA using the MMSE-STSA estimator and its generalized method as the post-processing part of BSSA. Also, we conduct a theoretical analysis for the amount of musical noise generation in the proposed method. From the theoretical analysis and objective evaluation results, we derive the optimal parameter for the proposed method.

Over the past decades the problem of one channel, speech enhancement has been addressed by a great deal of researchers. In this work selected methods belonging to a variety of categories are applied to denoise speech signals corrupted by non-stationary urban noise. The performance of spectral subtraction, signal subspace, model-based and Kalman filtering approaches is evaluated. Several objective measures which are designed to predict human listening tests are employed in order to reach accurate conclusions. Two series of experiments were carried out while multiband spectral subtraction along with a short-time spectral amplitude (STSA) estimator based on the minimization of the mean square error (MSE) of the log-spectra are shown to outperform the rest of the algorithms.

Event Abstract Back to Event Estimation of time-variant, multivariate AR-models with application to interaction analysis in the human brain Thomas Milde1*, Lutz Leistritz1, Thomas Weiss2 and Herbert Witte1 1 Friedrich Schiller University , Institute of Medical Statistics Computer Sciences and Documentation, Germany 2 Friedrich Schiller University Jena, Department of Psychology, Germany Aim:In this Study, a new Kalman-Filter approach is presented for the estimation of high dimensional time variant multivariate AR-processes. Methods: Firstly, the new Kalman filter algorithm and an ERP dataset derived from an experiment with noxious laser stimuli are presented. To validate the possibilities of the new method, the Kalman filter and the RLS algorithm with forgetting factor are compared with regard to their applicability to high-dimensional time series. In order to test both approaches simulated and measured time series were used. In a multi-trial approach directed interactions between event-related potentials (ERPs) were computed. The time-variant Granger Causality Index was used for interaction analysis. Results: It can be shown that the Kalman approach enables a time-variant parameter estimation of a 58-dimensional multivariate AR model. The RLS-based algorithm fails for dimensions higher than . The interaction analysis on the simulated data-set shows, that the Kalman-filter produces more reliable time-variant granger causality indices than the RLS. Conclusion: The high-dimensional AR model provides an improved neurophysiological interpretation of the computed interaction networks. Kalman filter based algorithms are applicable for such high-dimensional analysis. One of the advantages of the Kalman filter based algorithms is the multiplicity to design the course of the estimated time-variant AR-parameters. This is impossible with just one single forgetting factor like RLS provides. Acknowledgments: This study was in part supported by the Federal Ministry of Education and Research (Bernstein Group, 01GQ0703; application), the German Research Council (DFG Wi 1166/9-1 Gamma; method's development) and the COST Action BM0601 NEUROMATH. Conference: NeuroMath COST Action BM0601: Neurodynamic Insight into Functional Connectivity, Cognition, and Consciousness, Dubrovnik, Croatia, 27 Mar - 28 Mar, 2010. Presentation Type: Poster Presentation Topic: Posters Citation: Milde T, Leistritz L, Weiss T and Witte H (2010). Estimation of time-variant, multivariate AR-models with application to interaction analysis in the human brain. Front. Neurosci. Conference Abstract: NeuroMath COST Action BM0601: Neurodynamic Insight into Functional Connectivity, Cognition, and Consciousness. doi: 10.3389/conf.fnins.2010.05.00030 Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters. The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated. Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed. For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions. Received: 24 Jul 2010; Published Online: 24 Jul 2010. * Correspondence: Thomas Milde, Friedrich Schiller University, Institute of Medical Statistics Computer Sciences and Documentation, Jena, Germany, thomas.milde@mti.uni-jena.de Login Required This action requires you to be registered with Frontiers and logged in. To register or login click here. Abstract Info Abstract The Authors in Frontiers Thomas Milde Lutz Leistritz Thomas Weiss Herbert Witte Google Thomas Milde Lutz Leistritz Thomas Weiss Herbert Witte Google Scholar Thomas Milde Lutz Leistritz Thomas Weiss Herbert Witte PubMed Thomas Milde Lutz Leistritz Thomas Weiss Herbert Witte Related Article in Frontiers Google Scholar PubMed Abstract Close Back to top Javascript is disabled. Please enable Javascript in your browser settings in order to see all the content on this page.

Obstructive sleep apnea (OSA) is a highly prevalent disease in which upper airways are collapsed during sleep, leading to serious consequences. Snoring is the earliest symptom of OSA, but its potential in clinical diagnosis is not fully recognized yet. The first task in the automatic analysis of snore-related sounds (SRS) is to segment the SRS data as accurately as possible into three main classes: snoring (voiced non-silence), breathing (unvoiced non-silence) and silence. SRS data are generally contaminated with background noise. In this paper, we present classification performance of a new segmentation algorithm based on pattern recognition. We considered four features derived from SRS to classify samples of SRS into three classes. The features—number of zero crossings, energy of the signal, normalized autocorrelation coefficient at 1 ms delay and the first predictor coefficient of linear predictive coding (LPC) analysis—in combination were able to achieve a classification accuracy of 90.74% in classifying a set of test data. We also investigated the performance of the algorithm when three commonly used noise reduction (NR) techniques in speech processing—amplitude spectral subtraction (ASS), power spectral subtraction (PSS) and short time spectral amplitude (STSA) estimation—are used for noise reduction. We found that noise reduction together with a proper choice of features could improve the classification accuracy to 96.78%, making the automated analysis a possibility.

The paper describes a new speech enhancement approach which employs adaptive /spl beta/-order minimum mean square error (MMSE) spectral estimation of the short time spectral amplitude (STSA) of a speech signal. In the proposed approach, the human perceptual auditory masking effect is incorporated into the speech enhancement algorithm. The relationship between the value of /spl beta/ and the noise masking threshold is considered. The algorithm is based on a criterion by which the audible noise may be masked rather than being attenuated, thereby reducing the chance of speech distortion. Performance assessment is given to show that our proposal can achieve a more significant noise reduction and a better spectral estimation of weak speech spectral components from a noisy signal as compared to many existing speech enhancement algorithms.

In permanent magnet synchronous machine, high-frequency (HF) signal injection has been extensively investigated for permanent magnet temperature (PMT) estimation, in which PMT is estimated from the temperature-dependent HF resistance. Existing studies require prior knowledge on the HF resistance and neglect the fact that PMT is temporally correlated. This paper proposes a state-space model for PMT estimation, in which PMT is modeled with a piecewise linear equation to explore the temporal correlation. The state-space model is nonlinear due to unknown model parameters, which is required to be known in existing studies. This paper proposes to use expectation maximization particle filter (EM-PF) for simultaneous PMT and model parameter estimation. After EM-PF estimation, the state-space model becomes linear, so Kalman filter is employed for online PMT estimation. The proposed EM-PF along with a Kalman-filter-based approach can explore the temporal correlation among PMTs to improve the estimation performance, which can be hardly achieved in existing studies regarding PMT as a time-independent parameter. It should be noted that EM-PF is for initial PMT and model parameter estimation, while Kalman filter is for online PMT estimation ensuring computation efficiency and real-time capability. Our approach is validated with both numerical and experimental investigations.

In order to solve the problem of parameter estimation of nonlinear model in chemical engineering, a novel heuristic optimization approach is presented. The proposed approach develops and employs particle swarm optimization (PSO) technique to search for optimal solution of the parameter estimation. Firstly, the proposed approach converts the problem of parameter estimation of nonlinear model into an equation solving problem that is prone to be optimized. Secondly, two extrema of PSO are used as the target, and all the particles synchronously move through the solution space roundly and efficiently to find the optimum parameter estimation. The feasibility of the proposed method is demonstrated by the parameter estimation example of the pyrolysis model for fir.

In this paper, we provide a theoretical analysis of the minimum mean-square error short-time spectral amplitude (MMSE-STSA) estimator with the biased a priori SNR estimation and its extension to musical-noise-free speech enhancement. Recently, musical-noise-free speech enhancement has been proposed, where no musical noise is generated in iterative spectral subtraction. However, no existence of the musical-noise-free condition in the MMSE-STSA estimator has been reported. Therefore, in this paper, we show that the musical-noise-free condition exists in the biased MMSE-STSA estimator via the theoretical analysis. In addition, we perform comparative experiments and clarify the efficacy of the proposed musical-noise-free speech enhancement.

Due to the complexity of hydrodynamic forces during maneuvering, the parameter estimation of the hydrodynamic model of vessel is challenging. This paper proposed a simplified three-degree-of-freedom (3DOF) hydrodynamic model of a two-thruster vessel. Based on experiment data, a comparative study on Extended Kalman filter (EKF) and Unscented Kalman filter (UKF) performance in non-linear parameter estimation is conducted. Via the comparison of model accuracy, the feasibility and possible profits of UKF are demonstrated.

The strip Kalman filtering previously proposed by B.R. Suresh and B.A. Shenoi (ibid, vol.CAS-28, no.4, p.307-19, 1981) for image restoration is reconsidered. It is pointed out that the procedure given there for parameter estimation of the image model does not take into account the vector nature of the image process and as a result can lead to incorrect identification. It is also shown that for the composite dynamic model derived there the standard Kalman filtering equation cannot be applied, as the blur states in this model should be estimated one step ahead. >

A treatment system for reducing stuttering uses an auditory feedback modification technique to train the stutterer's speech motor control system to be more stable. The auditory feedback modification is based on a model of speech motor control in the human brain that incorporates a variation of observer-based control and Smith prediction. In addition, the Kalman gain of the model is set by comparing the speech muscular control signals sent to the person's vocal tract to the corresponding auditory speech sounds the person actually hears. It is believed that the speech motor control system of a stutterer has set the Kalman gain too high, thereby creating an unstable control system that in turn causes stuttering. A feedback modifier feeds back the stutterer's speech to the stutterer with a perturbation that is small enough to pass a validator function that is believed to be part of the speech motor control system. The small perturbations increase the difference between the target speech and the fed back speech, which is believed to cause the speech motor control system to decrease the Kalman gain. Over a treatment program, it is believed that the perturbations "train" or adapt the stutterer's speech motor control system so as to reduce the Kalman gain, which is further believed to persist when the auditory feedback is discontinued.