Non-stationary Noise Reduction Research Articles

A hidden Markov model method for non-stationary noise reduction: case study on Sentinel data for mowing detection

We propose a method for reducing the non-stationary noise in signal time series of Sentinel data, based on a hidden Markov model. Our method is applied on interferometric coherence from Sentinel-1 and the normalized difference vegetation index (NDVI) from Sentinel-2, for detecting the mowing events based on long short-term memory (LSTM). With integrating our noise reduction step to the LSTM neural network architecture, we improved the $$F_1$$ -score from 0.69 to 0.76.

Noise reduction method for non-stationary noise of vibration signal of shearer mechanical transmission system

Aiming at the problem that vibration signal of mechanical transmission system of shearer is vulnerable to be interferenced by non-stationary strong noise, a combined method with variational mode decomposition and recursive least squares (VMD-RLS) is proposed for non-stationary noise reduction. Firstly, the vibration signal is decomposed by VMD and the number of decomposition layers is determined according to the principle of minimum energy difference. Then, a dominant criterion of frequency-domain characteristic amplitude is proposed to select invalid intrinsic mode components (IMF, Intrinsic Mode Function). The proportion of the frequency-domain characteristic amplitude of each order IMF component for VMD is calculated. The invalid intrinsic mode functions are selected for reconstruction, and the noise signal after VMD decomposition is obtained. Finally, the noise signal decomposed by VMD is used as reference noise signal of RLS adaptive filtering algorithm to carry out adaptive noise reduction, and the final vibration signal after noise reduction is obtained. The wavelet de-noising method, VMD de-noising method, and VMD-RLS de-noising method are used to carry out noise reduction comparative experiment on the vibration signal of shearer. The experiment results indicate that the noise reduction method of VMD-RLS is better than VMD and wavelet de-noising method. The noise reduction method of VMD-RLS can effectively extract vibration signals from non-stationary strong noise, and provide help for the subsequent fault diagnosis and analysis of shearer mechanical transmission system.

Adaptive line enhancer for nonstationary harmonic noise reduction

In this paper, we propose a frequency-domain adaptive line enhancer (ALE) to reduce nonstationary harmonic noise, such as medical equipment beeps, from a noisy speech signal captured by a single microphone. The reduction of nonstationary noise is very challenging, with the tradeoff between noise reduction and speech distortion, often resulting with much noise residuals. The proposed ALE is a combination of the commonly-used forward adaptive linear filter and a non-causal backward adaptive linear filter used together with an indicator for the presence of transient noise. The proposed combined filter results in less noise residuals while preserving the speech components. We compare the proposed approach to conventional and recent methods, and show that it can outperform these methods, achieving lower distortion, more noise reduction, and overall better speech quality and intelligibility.

ЛОКАЛЬНО-АДАПТИВНАЯ ФИЛЬТРАЦИЯ НЕСТАЦИОНАРНОГО ШУМА В ДЛИТЕЛЬНЫХ ЭЛЕКТРОКАРДИОГРАФИЧЕСКИХ СИГНАЛАХ

The research subject of the article is the methods of locally adaptive filtering of non-stationary (from the point of view of its variance) noise in long-term electrocardiogram (ECG) signals. The goal is to develop locally adaptive algorithms for filtering noise with different a priori unknown levels of variance in real-time for ECG signals recorded with a standard sampling rate of 500 Hz. The tasks to be solved are: to investigate the effectiveness of the developed adaptive ECG filtering algorithms using numerical statistical estimates of processing quality in a wide range of additive Gaussian noise variance variation, to investigate the suppression of real non-stationary electromyographic (EMG) noise, and to analyze the application for normal and pathological ECG signals. The methods are integral and local indicators of the filter quality according to the criteria of the mean square error and the signal-to-noise ratio was obtained using numerical simulation (via Monte Carlo analysis). The following results were obtained: an adaptive method for real-time suppression of non-stationary noise in the ECG is proposed, the one-pass and the two-pass algorithms, and the algorithm with selective depending on the preliminary estimates of noise levels re-filtering have been developed on the method basis. Statistical estimates of the filters' efficiency and analysis of their outputs show a high degree of suppression of the noise with different levels of variance in the ECGs. The distortions absence while processing QRS-complex and high efficiency of suppression of Gaussian and real EMG noise with varying variance are demonstrated. The analysis of the output signals and plots of the local adaptation parameters and the adaptable parameters of the proposed algorithms confirms the high efficiency of filtering. The developed algorithms have been successfully tested for normal and pathological ECG signals. Conclusions. The scientific novelty of the results is the development of a locally adaptive method with noise and signal-dependent filter parameters switching and of the adaptive algorithms based on this method for non-stationary noise reduction in the ECG in real-time. This method does not require time for filter parameters adaptation and a priori information about the noise variance, and it has a high-speed performance in real-time mode.

Semi-supervised transient noise suppression using OMLSA and SNMF algorithms

Transient interferences such as keystrokes, mouse clicks and hammering pose a significant challenge in the single channel speech enhancement due to their abrupt and non-continuous nature. Traditional noise suppression algorithms and even many non-stationary noise reduction algorithms do not adequately suppress transient interference. Therefore, in this work, we propose a semi-supervised single channel transient noise suppression method to effectively suppress the transient interference without significant audible distortion. The proposed algorithm consists of training and testing stages. In the training stage, the proposed technique first uses the optimally modified-log spectral amplitude (OMLSA) estimator to estimate the transient noise from the noisy speech signal. After that, we eliminate the residual speech components from the estimated noise obtained from OMLSA based on the correlation coefficient, by taking correlation between the estimated noise with the available clean speech data from the dataset passed through the voice activity detector for silence zones removal. Afterwards, we use this noise for training the noise dictionary in sparse non-negative matrix factorization. Clean speech data is used for speech dictionary training. In the enhancement stage, the dictionaries are fixed and concatenated, to obtain the corresponding activation matrices. The clean speech dictionary and the corresponding weight matrix are used to reconstruct the estimated speech. The experimental results reveal that the proposed algorithm provided better performance compared to other existing algorithms in the speech quality evaluation metrics.

Combined adaptive beamforming schemes for nonstationary interfering noise reduction

This paper introduces new adaptive beamforming methods for nonstationary noise reduction, designed to be robust against broadband interfering signals. In particular, we propose combined beamforming schemes within a standard adaptive beamforming system, such as the generalized sidelobe canceller (GSC). The novelty of such combined adaptive beamformers relies on the use of different adaptive sidelobe cancelling structures which allow the system to achieve robustness in nonstationary noisy environments. The combined structures are based on the convex combination of two multiple-input single-output (MISO) adaptive systems with complementary capabilities. The whole beamformer benefits from such combination and results to be able to preserve the best properties of each system. We introduce two different adaptive schemes, whose difference lies in the way of combining the MISO systems. Moreover, we present a further adaptive beamforming scheme which generalizes the previous techniques, thus improving the robustness against nonstationary interfering signals in multisource environments. The effectiveness of the proposed systems is also assessed in a nonstationary dense multipath environment. The experiments show that the proposed combined beamforming schemes are capable of enhancing the desired signal even in the presence of nonstationary interfering signals.

Integrating recursive minimum tracking and codebook-based noise estimation for improved reduction of non-stationary noise

Conventional single-channel noise reduction algorithms typically have problems with non-stationary noise. Popular algorithms such as minimum statistics or voice-activity-detector-based methods rely on the assumption that the noise spectral characteristics change very slowly over time. Codebook-based approaches try to overcome this problem by incorporating a priori knowledge about speech and different noise types. These approaches perform a joint estimation of the speech and noise spectra on a frame-by-frame basis. The frames are typically 20–40 ms long so that fast fluctuations of the signal characteristics can be tracked instantaneously. However, these methods require a pitch estimator to prevent speech distortion as well as residual noise in voiced speech frames. In addition, they are not very robust against model mismatch. In this paper, we propose an integrated noise estimation algorithm that combines the ability of codebook-based algorithms to track non-stationary noise with the robustness of a recursive minimum-tracking-based noise estimation algorithm. An objective and subjective evaluation is provided. Results confirm the superiority of the proposed algorithm in non-stationary noise scenarios compared to state-of-the-art algorithms.

On-line Speech Enhancement by Time-Frequency Masking under Prior Knowledge of Source Location

This paper presents the source extraction system which can extract only target signals with constraints on source localization in on-line systems. The proposed system is a kind of methods for enhancing a target signal and suppressing other interference signals. But, the performance of proposed system is superior to any other methods and the extraction of target source is comparatively complete. The method has a beamforming concept and uses an improved time-frequency (TF) mask-based BSS algorithm to separate a target signal from multiple noise sources. The target sources are assumed to be in front and test data was recorded in a reverberant room. The experimental results of the proposed method was evaluated by the PESQ score of real-recording sentences and showed a noticeable speech enhancement. Keywords—Beamforming, Non-stationary noise reduction, Source separation, TF mask.

The combined effects of reverberation and fluctuating noise on sentence intelligibility

Listening conditions in everyday life typically include a combination of reverberation and nonstationary background noise. It is well known that sentence intelligibility is adversely affected by these factors. To assess their combined effects, a model is introduced that combines two models of speech perception, the Extended Speech Intelligibility Index (E-SII), and the Speech Transmission Index (STI). First, the effect of reverberation on nonstationary noise—reduction of modulations—is determined. The E-SII is then used to evaluate the effect of this modified nonstationary noise, while the STI is applied to quantify the effects of reverberation and noise on speech quality. To validate this model, speech reception thresholds (SRTs) were measured for ten normal-hearing listeners, under various combinations of nonstationary noise and artificially created reverberation. After taking the characteristics of the speech corpus into account, results show that the model accurately predicts SRTs in fluctuating noise and reverberation for normal-hearing listeners.

Neuro-rough control of masking thresholds for audio signal enhancement

The paper addresses the problem of neuro-rough hybridisation applied to digital processing of audio signals. Moreover, the application of some selected soft computing techniques to non-stationary noise reduction is described. Some attention is also put to a discussion of the intelligent decision algorithms performance. The noise reduction algorithm is based on the new perceptual approach exploiting some properties of the human auditory system. Furthermore, the paper introduces the engineered perceptual filter driven by an intelligent controller employing rules generated with the use of a rough set-based algorithm supported by a neural network. The goal of the intelligent controller is to estimate the current statistics of corrupting noise on the basis of the analysis of signals received from telecommunication channel. Thereafter, the noise estimate enables determining the masking threshold levels which allow making the noise inaudible in the audio signals. Since the implemented decision algorithm requires quantised data, thus the Kohonen's self-organising maps (SOM) extended by various distance metrics were used as data quantisers. Some results of the experiments in the domain of non-stationary noise reduction in speech are discussed in the paper.

Speech Recognition System using Wavelet Analysis. Reduction of Non-stationary Noise and Database Compression.

This paper presents a new speech recognition system using wavelet analysis. The framework of the system based on wavelet analysis is described with the efficiency and accuracy considered. The proposed system has following advantages comparing to conventional speech recognition systems based on short time Fourier transform : the non-stationary noise such as impulsive noise can be extracted and reduced, the database for speech recognition can be effectively compressed, and high recognition accuracy can be obtained for the burst speech which conventional systems have never been able to well recognize. Furthermore, the influence of data compression on recognition accuracy and recognition speed is examined.

Noise reduction in acoustic signals using the perceptual coding and intelligent decision systems

A new algorithm of broadband, nonstationary noise reduction was engineered employing the perceptual approach to the removal of noise from acoustic signals. It enables analysis and processing of sound according to characteristics of hearing sense, and employs a decision system to automatically adjust thresholds of masking. The presented algorithm provides an extension of perceptual coding applications, because it makes possible the reduction of noise included in source signals. Although noisy components of a signal may occur randomly, it is possible to estimate the noise distribution on the basis of signal analysis in silence passages of the transmission. Subsequently, the decision should be made as to the masking curves’ parameter settings which will make the noise inaudible. That is the reason why an intelligent approach is used in modeling some interrelations between available noise patterns and the noise affecting consecutive portions of useful signals. The decision system was implemented in various ways using fuzzy reasoning, rough set approach, and modified feedforward neural networks with functional links. The proposed methods of noise reduction and obtained results of noisy speech processing are presented in the paper. [Work supported by the Committee for Scientific Research, Poland, Grant No. 8 T11D 021 12.]

Reduction of nonstationary noise in telescope imagery using a support constraint

We demonstrate the use of image support constraints in a noise-reduction algorithm. Previous work has revealed serious limits to the use of support if image noise is wide-sense stationary in the frequency domain; we use simulation and numerical calculations to show these limits are removed for nonstationary noise generated by inverse-filtering adaptive optics image spectra. To quantify the noise reduction, we plot fractional noise removed by the proposed algorithm over a range of support sizes. We repeat this calculation for other noise sources with varying degrees of stationarity.