Speech enhancement using harmonics regeneration based on multiband excitation

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.

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.

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.

This paper deals with the problem of microphone array speech enhancement using a hybrid Generalized Sidelobe Canceller (GSC), Near-Field Super-Directive (NFSD) beamformer, and post-filter. In this research, we employ a near field compensation block before the blocking matrix (of the GSC) to prevent signal leakage in the reference noise and a Linear Constrained Minimum Variance (LCMV) beamformer instead of blocking matrix to generalize the appropriate performance of the system for different arrival directions of desired speech. We also consider the application of the post-filters on the beamformer output. A modified version of McCowan post-filter is presented by calculating coherence function from microphone-array inputs. Considering similar idea, we also propose a multichannel version of Short-Time Spectral Amplitude (STSA) post-filter. Our evaluations clearly demonstrate the effectiveness of the proposed modifications in the enhancement of the noisy speech.

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.

This paper presents a Kalman tracking approach to re-estimate clean spectral amplitude from noisy speech spectrum for re-synthesis based speech enhancement. The motivation of using Kalman filter and training is to exploit the temporal correlation between speech dynamics and to include prior knowledge of speech to improve the model parameter estimation in harmonic noise model (HNM) based speech enhancement system. The re-estimated harmonic amplitude is fitted into an analysis-synthesis framework to accomplish a more accurate HNM based re-synthesis. Objective evaluation results show the proposed method achieves significant improvement over various classical short-time spectral amplitude (STSA) based methods, especially in low signal-to-noise ratio 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.

Refinement of Enhanced Speech Using Hybrid-Median Filter and Harmonic Regeneration Approach

The speech signal in general is corrupted by noise and the noise signal does not affect the speech signal uniformly over the entire spectrum. An improved Wiener filtering method is proposed in this paper for reducing background noise from speech signal in colored noise environments. In view of nonlinear variation of human ear sensibility in frequency spectrum, nonlinear multi-band Bark scale frequency spacing approach is used. The cross-correlation between the speech and noise signal is considered in the proposed method to reduce colored noise. To overcome harmonic distortion introduced in enhanced speech, in the proposed method regenerate the suppressed harmonics are regenerated. Objective and subjective tests were carried out to demonstrate improvement in the perceptual quality of speeches by the proposed technique.

This paper focuses on two representative single-microphone noise reduction problems: speech distortion and musical noise. Many noise reduction methods have been proposed for each problem. Harmonic regeneration noise reduction (HRNR) was introduced for the improvements of speech distortion and the a priori signal-to-noise ratio (SNR) estimator. HRNR involves using a unique signal to regenerate harmonics, which had been eliminated. Musical-noise-free noise reduction based on the minimum-mean square error short-time spectral amplitude estimator (musical-noise-free MMSE-STSA estimator) has also been proposed. This method can suppress a noisy signal without generating musical noise by introducing a bias into the classical a priori SNR estimator. We propose a noise reduction method for addressing these problems simultaneously by improving the classical a priori SNR estimator. We investigated the behavior of the internal parameters for the proposed and conventional methods with regard to speech quality and show the effectiveness of the proposed method in terms of speech distortion and musical noise. We consider and discuss the relationship between the estimation accuracy of an a priori SNR and speech quality. Specifically, we consider the factors to improve speech quality in terms of biasing.

In this paper, we propose a robust speech enhancement algorithm for nonstationary noise environments, which comprises a multiband spectral subtraction (MBSS) speech estimator and a minima controlled recursive averaging (MCRA) noise estimation. A multiband approach is obtained by dividing the whole spectrum into multiple subbands and applying spectral subtraction independently in each band. The noise estimate exploit the observation that the noise signal typically has a nonuniform effect on the spectrum of speech and is given by averaging the past spectrum power value and using a time and frequency dependent smoothing factor that is calculated based on the speech presence probability in subbands. We test the performance of proposed algorithm in various nonstationary noise environments. The experiment results confirm the superiority of the MBSS and MCRA estimator. Outstanding speech enhancement is achieved, while avoiding the residual musical noise phenomena.

Results for improving the quality of narrowband CELP-coded speech by enhancing the pitch periodicity and by regenerating the high-band components of speech spectra are reported. Multiband excitation (MBE) analysis is applied to enhance the pitch periodicity by re-synthesizing the speech signal using a harmonic synthesizer. The high-band magnitude spectra are regenerated by matching to low-band spectra using a trained wideband spectral codebook. Information about the voiced/unvoiced (V/UV) excitation in the high-band is derived from a training procedure and recovered by using the matched low-band index. Simulation results indicate that the quality of the wideband enhanced speech is significantly improved over the narrowband CELP-coded speech.

This paper presents a variable-rate speech coding at 800 bps, 1.2 kbps and 2.4 kbps based on the algorithm of multi-band excitation (MBE). It takes the advantages of MBE algorithm in pitch estimation, UV decision and the estimation of spectral amplitude vector. The main difficulty of at the low bit rate of the MBE algorithm lies in an efficient quantization of the spectral amplitude vector. The technique of linear prediction (LP) is used, and the spectral amplitude vectors are approached by an all-pole model with formants to be enhanced. The techniques of vector quantization (VQ) is adopted to quantize the LP parameters. The split-VQ (SPVQ), and multi-stage VQ (MSVQ) are employed to reduce the storage capacity and computing complexity, and max-min method is used to accelerate the searching of the VQ codebook. The 800 bps, 1.2 kbps and 2.4 kbps vocoder has been implemented in real-time using a single fixed point DSP ADSP2181 chip. The synthesized speech of this vocoder is clear and intelligible even in the presence of background noise at 80 dB. The speech quality of this vocoder has reached the current international level.

The speech enhancement effect of traditional deep learning algorithms is not ideal under low signal-to-noise ratios (SNR). Skip connections-deep neural network (Skip-DNN) improves the traditional deep neural network (DNN) by adding skip connections between each layer of the neural network to solve the degradation problem of DNN. In this paper, the Multiresolution Cochleagram (MRCG) features in the gammachirp transform domain are denoised to obtain the improved MRCG (I-MRCG). The noise reduction method adopts the Minimum Mean-Square Error Short-Time Spectral Amplitude Estimator (MMSE-STSA) and takes I-MRCG as the input feature and Skip-DNN as the training network to improve the speech enhancement effect of the model. This paper also proposes an improved source-to-distortion ratio (SDR) loss function. When the loss function uses the improved SDR, it will improve the performance of Skip-DNN speech enhancement model. The experiments in this paper are performed on the Edinburgh dataset. When using I-MRCG as the input feature of Skip-DNN, the average perceptual evaluation of speech quality (PESQ) is 2.9137, and the average short-time objective intelligibility (STOI) is 0.8515. Compared with MRCG as Skip-DNN input features, the improvements are 0.91% and 0.71%, respectively. When the improved SDR is used as the loss function of the speech model, the average PESQ is 2.9699 and the average STOI is 0.8547. Compared with other loss functions, the improved SDR has a better enhancement effect when used as the loss function of the speech enhancement model.

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.