Harmonic Phase Estimation in Single-Channel Speech Enhancement Using Phase Decomposition and SNR Information

The complexity of the speech enhancement problem has motivated many different solutions. However, most techniques address situations in which the target speech is fully intelligible and the background noise energy is low in comparison with that of the speech. Thus while current enhancement algorithms can improve the perceived quality, the intelligibility of the speech is not increased significantly and may even be reduced. Recent research shows that intelligibility of very noisy speech can be improved by the use of a binary mask, in which a binary weight is applied to each time-frequency bin of the input spectrogram. There are several alternative goals for the binary mask estimator, based either on the Signalto-Noise Ratio (SNR) of each time-frequency bin or on the speech signal characteristics alone. Our approach to the binary mask estimation problem aims to preserve the important speech cues independently of the noise present by identifying time-frequency regions that contain significant speech energy. The speech power spectrum varies greatly for different types of speech sound. The energy of voiced speech sounds is concentrated in the harmonics of the fundamental frequency while that of unvoiced sounds is, in contrast, distributed across a broad range of frequencies. To identify the presence of speech energy in a noisy speech signal we have therefore developed two detection algorithms. The first is a robust algorithm that identifies voiced speech segments and estimates their fundamental frequency. The second detects the presence of sibilants and estimates their energy distribution. In addition, we have developed a robust algorithm to estimate the active level of the speech. The outputs of these algorithms are combined with other features estimated from the noisy speech to form the input to a classifier which estimates a mask that accurately reflects the time-frequency distribution of speech energy even at low SNR levels. We evaluate a mask-based speech enhancer on a range of speech and noise signals and demonstrate a consistent increase in an objective intelligibility measure with respect to noisy speech.

Most algorithms for speech enhancement in the spectral domain focus on the acquisition of an estimator of the clean speech parameter, such as spectrum or amplitude. Enhanced speech quality with these methods mainly depends upon the accuracy of the estimator. When the signal-to-noise ratio (SNR) becomes lower, e.g., less than 3 dB, the enhanced speech often shows unsatisfactory quality. Here, we propose a new method for speech enhancement in the spectral domain. It is based on our view that speech characteristics are perceived mainly by part of the spectral components which have higher local (or instantaneous) SNRs. These components have a special importance on speech SNR enhancement and speech understanding. Identifying these components in the noisy spectrum with a decision system and constructing enhanced speech just by the same noisy amplitude and phase, the result is similar to those popular algorithms. All that is needed is a local SNR decision for each spectral component. The threshold for the decision is derived and two popular methods for local SNR (LSNR) estimation are suggested. It shows better performance under lower SNR signals. Though the number of these kind of spectral components varies with different signal SNR, any algorithms which can pick up more than 90% of it is sufficient to match any existing methods. Primary considerations and results are shown.

In single-channel speech enhancement the spectral amplitude of the noisy signal is often modified while the noisy spectral phase is directly employed for signal reconstruction. Recently, additional improvement in speech enhancement performance has been reported when the noisy phase is modified. In this work, we propose a Bayesian estimator for phase of harmonics given the noisy speech. The proposed estimator relies on the fundamental frequency and the signal-to-noise ratio at harmonics. Throughout our experiments, we evaluate the performance of the proposed phase enhancement in comparison with the noisy phase, a benchmark and the clean phase as the upper-bound. The proposed method leads to joint improvement in quality and intelligibility at different SNRs and noise types.

This study proposes a bi-directional recurrent neural network (Bi-RNN) post-processing method for speech enhancement (SE) at low signal-to noise ratios (SNR). Current speech enhancement solutions performed badly under low SNR situations. Loizou and Kim proposed a solution to reduce speech distortion errors in time-frequency (T-F) domain but it requires the knowledge of ground truth. As ground truth is unknown in real-life applications, the current study proposes to use a Bi-RNN to implement Loizou and Kim’s solution as a post-processing method for SE engines. Our solutions do not require prior knowledge of ground truth. The effectiveness of the proposed method is investigated with a spectral subtraction (SS) SE engine, a non-negative matrix factorization (NMF) SE engine, and a deep neural network ideal ratio mask (DNN-IRM) SE engine, under matched/mis-matched noise and different SNR conditions. Experimental results demonstrate that the proposed post-processing method effectively improved both perceptual evaluation of speech quality (PESQ) and short-time objective intelligibility (STOI) for all of these SE engines, especially at low SNR conditions.

This paper introduces novel two-channel a priori Signal-to-Noise Ratio (SNR) estimators for use in frequency-domain speech enhancement algorithms. The SNR estimation is based on statistics of the noisy phase difference between two microphones in each frequency bin. Namely, the corresponding probability distribution is derived assuming a complex Gaussian model, and is written in terms of the SNR only. This shifts the two-channel SNR estimation problem into a classical statistical parameter estimation problem, which we propose to solve via an online version of the Method of Moments. The usefulness of the resulting approach is demonstrated in a realistic speech enhancement experiment, where a simple Wiener-based solution is objectively found to perform competitively when compared with other recent two-channel noise reduction techniques.

In this article, the authors study the convergence of an iterative algorithm for the joint estimation of the signal-to-noise ratio (SNR) and the transmission channel in orthogonal frequency division multiplexing context. At each step of the algorithm, the authors use the minimum-mean-square error (MMSE)-based SNR estimation, which feeds the linear MMSE channel estimation. Reciprocally, this efficient channel estimation is used to perform the SNR estimation. The authors provide a proof of convergence of the algorithm to a single value. Furthermore, we derive an accurate approximation of the bias of the estimation. Simulations show that the algorithm converges quickly and verifies the theoretical results. They also show the efficiency of both SNR and channel estimation. By comparing with the existing methods, the authors show that the tradeoff between the number of required pilots in the preamble and the performance of the SNR estimation were improved. Furthermore, for a fixed bit error rate, the SNR gap between the proposed channel estimation and the perfect one is <0.5 dB.

Traditional speech enhancement algorithms have been widely studied due to simplicity and convenience, especially spectral subtraction algorithm. However, traditional speech enhancement algorithms always utilize the available noisy speech phase spectrum as the optimal estimation of clean speech phase spectrum, which is inaccurate at low signal-noise ratio (SNR). Aiming at improving performance of typical spectral subtraction speech enhancement algorithm under the circumstances of low SNR, a multiband spectral subtraction speech enhancement algorithm with phase spectrum compensation is proposed in this paper. The proposed method compensates noisy speech phase spectrum based on the SNR and estimation of noise obtained in the current frame and reduces musical noise generated by spectral subtraction. Simulation results show that the proposed algorithm has a better performance than traditional algorithms, especially at low SNR.

In this paper we have estimated signal to noise ratio (SNR) of active users in uplink scenario of CDMA systems blindly. We have not any knowledge about PN (pseudo noise) code, frequency, and timing of received signals. In this way first we synchronized received signal of desired user and then we use subspace method for SNR estimation of synchronized user. This type of SNR estimation is classified as non data aided (NDA) methods. Our synchronization is based on maximizing of covariance matrix Frobenius norm. After that by estimating PN code of strongest signal we are able to estimate transmitted signal data, and by using estimated SNR, we can eliminate this signal from received signal by using SIC (serial interference canceller). By repetition this cycle for each user we can estimate SNR of active users in uplink scenario. Also, in this paper we use minimum description length (MDL) criteria for estimating signal subspace dimension. Standard divisions of estimated SNR show that good agreement with exact SNR for all users.

In the channel-varying environment, it is very important to estimate the signal to noise ratio (SNR) of received signal and to transmit the signal effectively for the modern communication system. The performance of existing non-data-aided SNR estimation methods are substantially degraded for high level modulation scheme such as M-ary amplitude and phase shift keying or quadrature amplitude modulation. In this paper, we propose a SNR estimation method which uses zero point auto-correlation of received signal per block and auto/cross-correlation of decision feedback signal in orthogonal frequency division multiplexing (OFDM) system. Proposed method can be studied into two types; Type 1 can estimate SNR by zero point auto-correlation of decision feedback signal based on the second moment property. Type 2 uses both zero point auto-correlation and cross-correlation based on the fourth moment property. In block-by-block reception of OFDM system, these two SNR estimation methods can be possible for the practical implementation due to correlation based the estimation method and they show more stable estimation performance than the previous SNR estimation methods. Also, we mathematically derive the SNR estimation expression according to computational difference of auto/cross-correlation. Finally, Monte Carlo simulations are used to verify the proposed method.

&lt;p&gt;&lt;b&gt;Speech enhancement was, is, and will be the key technology for digital speech transmission. When developing speech enhancement algorithms, the intelligibility or the quality of the processed speech needs to be evaluated. Intelligibility is more fundamental than quality. The evaluation of speech intelligibility can be carried out through subjective listening tests and objective metrics. Carrying out subjective listening tests is time-consuming and costly. Using the objective metrics is more efficient. Thus, there has been an increasing research in speech intelligibility prediction.&lt;/b&gt;&lt;/p&gt; &lt;p&gt; Speech intelligibility can be measured in terms of the information received by a listener. This thesis aims at developing a mutual information-based speech intelligibility predictor (SIP) and using the mutual information-based predictor to assist the fitting process in hearing instruments. To achieve this goal, this thesis carried out three studies. &lt;/p&gt; &lt;p&gt;First, it studied the modeling of the transmitted message. For mutual information-based SIPs, the most important thing is to determine the transmitted message. This thesis studied two approaches of modeling of the message: one is the continuous-valued sound, and the other is the discrete-valued linguistic message. Two corresponding SIPs were developed. By comparing their predicted intelligibility results with the psychometric curves, which are the subjective intelligibility scores, it shows that the modeling of discrete-valued message gives a better match to the psychometric curves.&lt;/p&gt; &lt;p&gt; Second, based on the modeling of the discrete-valued message, this thesis proposed a mutual information-based SIP. Since the discrete-valued message cannot be obtained from a speech signal, the proposed SIP calculates the mutual information between the clean speech and the received speech, instead of calculating the mutual information between the message and the received speech. The proposed SIP considers frequency correlation for both the clean speech and the received speech. The evaluation results show that the proposed SIP performs better than the existing state-of-the-art mutual information-based SIPs.&lt;/p&gt; &lt;p&gt;Third, this thesis proposed an automatic fitting tool for the nonlinear frequency compression (NFC) operator, which is a frequency lowering operator used in hearing instruments. The automatic fitting tool adjusts the parameter in the NFC by maximizing the mutual information between the message and the frequency-lowered speech. To evaluate the automatic fitting tool, the parameter was also searched by listening tests. The results show that the parameter determined by the automatic fitting tool is consistent to the parameter determined by the listening tests.&lt;/p&gt;

This paper deals with the problem of estimating the average signal-to-noise ratio (SNR) for a communication system employing non-coherent binary frequency shift keying (NCBFSK) over fading channels with white Gaussian noise (AWGN). The maximum likelihood (ML) estimator and one using data statistics have been derived and simulated for various scenarios including data-aided (DA), non-data aided (NDA) and joint estimation using both the data and pilot sequences. We also derive the Cramer-Rao bound (CRB) for the estimators. The results show that for a particular region of interest (e.g. high SNR or low SNR) and depending upon the availability of pilot sequence, a particular SNR estimation scheme is suitable.

This work is part of an on-going effort to implement real-time biosignal quality analysis for electrocardiograms (ECG). Biosignal quality analysis can provide an estimate of the signal-to-noise ratio (SNR), which can be used to gate alarms for myocardial ischemia, as indicated by ST deviations in ambulatory ECG. Currently, the false alarm rate is high due to contaminants in the ECG, such as motion artifact. In the proposed algorithm, ECG data are segmented into 16-beat analysis windows. Within an analysis window, the heart's PQRST complexes are segmented, aligned, and averaged to form an estimate of the true PQRST complex output by the heart. A SNR is estimated for each beat by comparing the PQRST complex of each beat to the average PQRST complex; the lowest SNR across all beats in the analysis window is the SNR assigned to the analysis window. Performance of the algorithm is evaluated using ECG data contaminated with various levels of motion artifact. The algorithm provides SNR estimates that are correlated with true SNR (r = 0.89). Leveraging the repeatability of the PQRST complex, it derives these estimates solely from the recorded ECG and does not require a noise free period or any supplementary signals. Results indicate that this algorithm may be a viable method for gating myocardial alarms for reduction of the false alarm rate.

In this paper, we considered the problem of the speech enhancement similar to the real-world environments where several complex noise sources simultaneously degrade the quality and intelligibility of a target speech. The existing literature on the speech enhancement principally focuses on the presence of one noise source in mixture signals. However, in real-world situations, we generally face and attempt to improve the quality and intelligibility of speech where various complex stationary and nonstationary noise sources are simultaneously mixed with the target speech. Here, we have used deep learning for speech enhancement in complex-noisy environments and used ideal binary mask (IBM) as a binary classification function by using deep neural networks (DNNs). IBM is used as a target function during training and the trained DNNs are used to estimate IBM during enhancement stage. The estimated target function is then applied to the complex-noisy mixtures to obtain the target speech. The mean square error (MSE) is used as an objective cost function at various epochs. The experimental results at different input signal-to-noise ratio (SNR) showed that DNN-based complex-noisy speech enhancement outperformed the competing methods in terms of speech quality by using perceptual evaluation of speech quality (PESQ), segmental signal-to-noise ratio (SNRSeg), log-likelihood ratio (LLR), weighted spectral slope (WSS). Moreover, short-time objective intelligibility (STOI) reinforced the better speech intelligibility.

After speech enhancement, some noisy speech will lose some effective speech information and cause speech signal distortion while improving its signal-to-noise ratio (SNR). According to the auditory masking effect, a method for judging the noise tolerance of noisy speech is proposed in this paper. Firstly, the necessity of judging noise tolerance is verified and analyzed. Then, based on the minimum priori SNR, the noise tolerance of noisy speech is studied, and its judgment threshold is given. Before speech enhancement, the voice segment is prejudged based on this threshold. If the minimum priori SNR ratio is less than this value at each frequency point, it is considered that the noise is intolerable at this time, and the speech enhancement is necessary. If the contrary is the case, it is determined that the noise is tolerable, and it is not necessary for speech enhancement. Finally, experiments are carried out to verify the effectiveness of the proposed method in this paper so that speech with tolerable noise can avoid speech distortion caused by enhancement.

In wireless orthogonal frequency division multiplexing (OFDM) systems, the knowledge of signal-to-noise ratio (SNR) plays an important role for system optimization. Most of the exiting literatures have studied the SNR estimation for perfect synchronization in additive white Gaussian noise (AWGN) channels, or in frequency selective channels. However, the realistic channels are always doubly selective, and there may exist some residual symbol timing offset (STO) and carrier frequency offset (CFO) after coarse synchronization in the time domain. In this paper, we take these into account and propose a time-domain average SNR estimation scheme using one OFDM training symbol (preamble) which has been divided into multiple parts with equal length. Compared with the timedomain low-complexity SNR estimator (TLSE), our proposed time-domain preamble-based SNR estimator (TPSE) can not only be implemented in the presence of STO and CFO, but also generate more accurate SNR estimation for the channels with large Doppler shift. The estimated SNR value based on our scheme in the time-domain can be further reused to improve the ensuing CFO estimation and fine synchronization efficiently.