Speech enhancement aims to improve the quality and intelligibility of speech using various techniques and algorithms. The speech signal is always accompanied by background noise. The speech and communication processing systems must apply effective noise reduction techniques in order to extract the desired speech signal from its corrupted speech signal. In this project we study wavelet and wavelet transform, and the possibility of its employment in the processing and analysis of the speech signal in order to enhance the signal and remove noise of it. We will present different algorithms that depend on the wavelet transform and the mechanism to apply them in order to get rid of noise in the speech, and compare the results of the application of these algorithms with some traditional algorithms that are used to enhance the speech. The basic principles of the wavelike transform are presented as an alternative to the Fourier transform. Or immediate switching of the window The practical results obtained are based on processing a large database dedicated to speech bookmarks polluted with various noises in many SNRs. This article tends to be an extension of practical research to improve speech signal for hearing aid purposes. Also learn about the main frequency of letters and their uses in intelligent systems, such as voice control systems.

The speech enhancement techniques are often employed to improve the quality and intelligibility of the noisy speech signals. This paper discusses a novel technique for speech enhancement which is based on Singular Value Decomposition. This implementation utilizes a Genetic Algorithm based optimization method for reducing the effects of environmental noises from the singular vectors as well as the singular values of a noise-corrupted speech. The presented article also reviews the existing algorithms for subspace division and carries out extensive sets of experiments to clearly show the efficiency of the proposed method in comparison with the other superior speech enhancement approaches.

One of the most widely used beamforming algorithms for the application of speech enhancement is the Minimum Variance Distortionless Response (MVDR) technique. The optimal coefficients of the MVDR beamformer are calculated based on the incoherence assumption of environmental interferences and the desired signal. Due to the nature of noise and speech signals, this assumption is not valid in many practical situations. This, in turn, results in inaccurateness of derived coefficients of the MVDR. In this paper, as the first change in the MVDR beamformer, by applying the eigenvalue analysis to the desired signal covariance matrix and removing small eigenvalues, the accuracy of the beamformer coefficients is improved. As the second contribution, we use a generalized version of the Short-Time Fourier Transform (STFT), namely the Short-Time Fractional Fourier Transform (STFrFT), to calculate the MVDR beamformer weights. In this research, after obtaining the optimal value of STFrFT parameter experimentally, the effect of each of the above two changes on the performance is investigated and compared with the basic methods. The results show that the proposed methods, while being stable to the changes of parameters and environmental conditions, achieve signal-to-noise ratio (SNR) values between  and , while the performance of the baseline method is in the range of . Although each of the above changes alone improves the performance, it is noted that the superior performance is obtained when both changes are applied together on the beamformer.

In this paper, a speech enhancement method based on sparse representation of data frames has been presented. Speech enhancement is one of the most applicable areas in different signal processing fields. The objective of a speech enhancement system is improvement of either intelligibility or quality of the speech signals. This process is carried out using the speech signal processing techniques to attenuate the background noise without causing any distortion in the speech signal. In this paper, we focus on the single channel speech enhancement corrupted by the additive Gaussian noise. In recent years, there has been an increasing interest in employing sparse representation techniques for speech enhancement. Sparse representation technique makes it possible to show the major information about the speech signal based on a smaller dimension of the original spatial bases. The capability of a sparse decomposition method depends on the learned dictionary and matching between the dictionary atoms and the signal features. An over complete dictionary is yielded based on two main steps: dictionary learning process and sparse coding technique. In dictionary selection step, a pre-defined dictionary such as the Fourier basis, wavelet basis or discrete cosine basis is employed. Also, a redundant dictionary can be constructed after a learning process that is often based on the alternating optimization strategies. In sparse coding step, the dictionary is fixed and a sparse coefficient matrix with the low approximation error has been earned. The goal of this paper is to investigate the role of data-based dictionary learning technique in the speech enhancement process in the presence of white Gaussian noise. The dictionary learning method in this paper is based on the greedy adaptive algorithm as a data-based technique for dictionary learning. The dictionary atoms are learned using the proposed algorithm according to the data frames taken from the speech signals, so the atoms contain the structure of the input frames. The atoms in this approach are learned directly from the training data using the norm-based sparsity measure to earn more matching between the data frames and the dictionary atoms. The proposed sparsity measure in this paper is based on Gini parameter. We present a new sparsity index using Gini coefficients in the greedy adaptive dictionary learning algorithm. These coefficients are set to find the atoms with more sparsity in the comparison with the other sparsity indices defined based on the norm of speech frames. The proposed learning method iteratively extracts the speech frames with minimum sparsity index according to the mentioned measures and adds the extracted atoms to the dictionary matrix. Also, the range of the sparsity parameter is selected based on the initial silent frames of speech signal in order to make a desired dictionary. It means that a speech frame of input data matrix can add to the first columns of the over complete dictionary when it has not a similar structure with the noise frames. The data-based dictionary learning process makes the algorithm faster than the other dictionary learning methods for example K-singular value decomposition (K-SVD), method of optimal directions (MOD) and other optimization-based strategies. The sparsity of an input frame is measured using Gini-based index that includes smaller measured values for speech frames because of their sparse content. On the other hand, high values of this parameter can be yielded for a frame involved the Gaussian noise structure. The performance of the proposed method is evaluated using different measures such as improvement in signal-to-noise ratio (ISNR), the time-frequency representation of atoms and PESQ scores. The proposed approach results in a significant reduction of the background noise in comparison with other dictionary learning methods such as principal component analysis (PCA) and the norm-based learning method that are traditional procedures in this context. We have found good results about the reconstruction error in the signal approximations for the proposed speech enhancement method. Also, the proposed approach leads to the proper computation time that is a prominent factor in dictionary learning methods.

Most real-time speech signals are frequently disrupted by noise such as traffic, babbling, and background noises, among other things. The goal of speech denoising is to extract the clean speech signal from as many distorted components as possible. For speech denoising, many researchers worked on sparse representation and dictionary learning algorithms. These algorithms, however, have many disadvantages, including being overcomplete, computationally expensive, and susceptible to orthogonality restrictions, as well as a lack of arithmetic precision due to the usage of double-precision. We propose a greedy technique for dictionary learning with sparse representation to overcome these concerns. In this technique, the input signal's singular value decomposition is used to exploit orthogonality, and here the ℓ1-ℓ2 norm is employed to obtain sparsity to learn the dictionary. It improves dictionary learning by overcoming the orthogonality constraint, the three-sigma rule-based number of iterations, and the overcomplete nature. And this technique has resulted in improved performance as well as reduced computing complexity. With a bit-precision of Q7 fixed-point arithmetic, this approach is also used in resource-constrained embedded systems, and the performance is considerably better than other algorithms. The greedy approach outperforms the other two in terms of SNR, Short-Time Objective Intelligibility, and computing time.

In this paper, a modified multi band spectral subtraction is proposed for speech denoising. In this method by estimating the non-stationary noise statistics an estimate of the SNR for the speech signal is obtained which is used in the adaptive averaging of the speech signal frames. In addition, by employing such estimate an over-subtraction factor can be obtained which substantially reduces the musical noise. The adjustment of the final algorithm parameters is accomplished based on the obtained SNR, such that the final algorithm has a minimum musical noise and distortion in the speech signal. The obtained results from the Itakura-Saito, global SNR and segmental SNR show that the proposed algorithm has an output signal with a superior quality compared to the multi-band spectral subtraction algorithm and its modified version.

In this paper, genetic programming is applied for quality improvement of noisy speech signal. Therefore, a system including both spectral subtraction and genetic programming is implemented for speech enhancement. In the proposed method, first noise is reduced by spectral subtraction. In the next step, genetic programming trees are trained for more enhancement of noisy signal by mapping the signal obtained by spectral subtraction to clean data. The proposed hybrid method improves signal to noise ratio about 2 to 6.5 dB. Comparison of genetic programming, multi-layer perceptron neural network, spectral subtraction, and the proposed hybrid method for speech enhancement indicates that the combination of spectral subtraction and genetic programming presents much better quality for enhanced signal compared to the other methods studied in this paper.

The short-and the long-term information in speech signal are useful for speech enhancement, especially if the speech signal is corrupted by both stationary and non-stationary noises. This paper proposes a new approach to provide long-term speech input for a deep denoising autoencoder by reducing the number of frequency sub-bands of the input data. This paper also proposes a two phase speech enhancement approach. The first phase performs short-term speech enhancement by using a deep denoising autoencoder. In the second phase, long-term speech enhancement denoising autoencoder is applied on the output of short-term enhanced speech data. The proposed models were evaluated on the Aurora-2 Speech recognition corpus and our results show significant improvements of 0. 3 in PESQ score at lower SNR values. The proposed models were evaluated on the recognition task where the proposed method results in 4% reduction in word error rate for the multi-condition training when compared to the baseline MFCC front-end.