Marcin Ciolek

Elimination of Impulsive Disturbances From Archive Audio Signals Using Bidirectional Processing

In this application-oriented paper we consider the problem of elimination of impulsive disturbances, such as clicks, pops and record scratches, from archive audio recordings. The proposed approach is based on bidirectional processing-noise pulses are localized by combining the results of forward-time and backward-time signal analysis. Based on the results of specially designed empirical tests (rather than on the results of theoretical analysis), incorporating real audio files corrupted by real impulsive disturbances, we work out a set of local, case-dependent fusion rules that can be used to combine forward and backward detection alarms. This allows us to localize noise pulses more accurately and more reliably, yielding noticeable performance improvements, compared to the traditional methods, based on unidirectional processing. The proposed approach is carefully validated using both artificially corrupted audio files and real archive gramophone recordings.

Automated Detection of Sleep Apnea and Hypopnea Events Based on Robust Airflow Envelope Tracking in the Presence of Breathing Artifacts

The paper presents a new approach to detection of apnea/hypopnea events, in the presence of artifacts and breathing irregularities, from a single-channel airflow record. The proposed algorithm identifies segments of signal affected by a high amplitude modulation corresponding to apnea/hypopnea events. It is shown that a robust airflow envelope-free of breathing artifacts-improves effectiveness of the diagnostic process and allows one to localize the beginning and the end of each episode more accurately. The performance of the approach, evaluated on 30 overnight polysomnographic recordings, was assessed using diagnostic measures such as accuracy, sensitivity, specificity, and Cohens coefficient of agreement; achieving 95%, 90%, 96%, and 0.82, respectively.

Elimination of impulsive disturbances from stereo audio recordings using vector autoregressive modeling and variable-order Kalman filtering

This paper presents a new approach to elimination of impulsive disturbances from stereo audio recordings. The proposed solution is based on vector autoregressive modeling of audio signals. On-line tracking of signal model parameters is performed using the stability-preserving Whittle-Wiggins-Robinson algorithm with exponential data weighting. Detection of noise pulses and model-based interpolation of the irrevocably distorted samples is realized using an adaptive, variable-order Kalman filter. The proposed approach is evaluated on a set of clean audio signals contaminated with real click waveforms extracted from silent parts of old gramophone recordings.

Akaike's final prediction error criterion revisited

When local identification of a nonstationary ARX system is carried out, two important decisions must be taken. First, one should decide upon the number of estimated parameters, i.e., on the model order. Second, one should choose the appropriate estimation bandwidth, related to the (effective) number of input-output data samples that will be used for identification/tracking purposes. Failure to make the right decisions results in the model deterioration, both in the quantitative and qualitative sense. In this paper, we show that both problems can be solved using the suitably modified Akaikes final prediction error criterion. The proposed solution is next compared with another one, based on the Rissanens predictive least squares principle.

On adaptive covariance and spectrum estimation of locally stationary multivariate processes

Abstract When estimating the correlation/spectral structure of a locally stationary process, one has to make two important decisions. First, one should choose the so-called estimation bandwidth, inversely proportional to the effective width of the local analysis window, in the way that complies with the degree of signal nonstationarity. Too small bandwidth may result in an excessive estimation bias, while too large bandwidth may cause excessive estimation variance. Second, but equally important, one should choose the appropriate order of the spectral representation of the signal so as to correctly model its resonant structure–when the order is too small, the estimated spectrum may not reveal some important signal components (resonances), and when it is too high, it may indicate the presence of some nonexistent components. When the analyzed signal is not stationary, with a possibly time-varying degree of nonstationarity, both the bandwidth and order parameters should be adjusted in an adaptive fashion. The paper presents and compares three approaches allowing for unified treatment of the problem of adaptive bandwidth and order selection for the purpose of identification of nonstationary vector autoregressive processes: the cross-validation approach, the full cross-validation approach, and the approach that incorporates the multivariate version of the generalized Akaike’s final prediction error criterion. It is shown that the latter solution yields the best results and, at the same time, is very attractive from the computational viewpoint.

On adaptive selection of estimation bandwidth for analysis of locally stationary multivariate processes

When estimating the correlation/spectral structure of a locally stationary process, one should choose the so-called estimation bandwidth, related to the effective width of the local analysis window. The choice should comply with the degree of signal nonstationarity. Too small bandwidth may result in an excessive estimation bias, while too large bandwidth may cause excessive estimation variance. The paper presents a novel method of adaptive bandwidth selection. The proposed approach is based on minimization of the cross-validatory performance measure for a local vector autoregressive signal model and, unlike the currently available methods, does not require assignment of any user-dependent decision thresholds.

Localization of impulsive disturbances in archive audio signals using predictive matched filtering

The problem of elimination of impulsive disturbances from archive audio signals is considered and its new solution, called predictive matched filtering, is proposed. The new approach is based on the observation that a large percentage of noise pulses corrupting archive audio recordings have highly repetitive shapes that match several typical “patterns”, called click templates. To localize noise pulses, click templates can be correlated with the sequence of multi-step-ahead prediction errors yielded by the model-based signal predictor. It is shown that predictive matched filtering is an efficient and computationally affordable disturbance localization technique - when combined with the classical detection method based on autoregressive modeling, it can significantly improve restoration results.

Renovation of archive audio recordings using sparse autoregressive modeling and bidirectional processing

The paper presents a new approach to elimination of broadband noise and impulsive disturbances from archive audio recordings. The proposed adaptive Kalman-like algorithm, based on a sparse autoregressive model of the audio signal, simultaneously detects noise pulses, interpolates the irrevocably distorted samples and performs signal smoothing. It is shown that bidirectional (forward-backward) processing of the archive signal improves smoothing efficiency and allows one to localize noise pulses more accurately, leading to noticeable performance improvements compared to unidirectional processing.

Identification of nonstationary multivariate autoregressive processes – Comparison of competitive and collaborative strategies for joint selection of estimation bandwidth and model order

Abstract The problem of identification of multivariate autoregressive processes (systems or signals) with unknown and possibly time-varying model order and time-varying rate of parameter variation is considered and solved using parallel estimation approach. Under this approach, several local estimation algorithms, with different order and bandwidth settings, are run simultaneously and compared based on their predictive performance. First, the competitive decision schemes are considered. It is shown that the best parameter tracking results can be obtained when the order is selected based on minimization of the appropriately modified Akaikes final prediction error statistic, and the bandwidth is chosen using the localized version of the Rissanens predictive least squares statistic. Next, it is shown that estimation results can be further improved if a collaborative decision is made by means of applying the Bayesian model averaging technique.

Detection of impulsive disturbances in archive audio signals

In this paper the problem of detection of impulsive disturbances in archive audio signals is considered. It is shown that semi-causal/noncausal solutions based on joint evaluation of signal prediction errors and leave-one-out signal interpolation errors, allow one to noticeably improve detection results compared to the prediction-only based solutions. The proposed approaches are evaluated on a set of clean audio signals contaminated with real click waveforms extracted from silent parts of old gramophone recordings.