A study on single-channel non-stationary noise suppression for cardiac sound

Over the decades, the noise-suppression (NS) methods for speech enhancement (SE) have been widely utilized, including the conventional signal processing methods and the deep neural networks (DNN) methods. Although stationary-noise can be suppressed successfully using conventional or DNN methods, it is significantly challenging while suppressing the non-stationary noise, especially the transient noise. Compared to conventional NS methods, DNN NS methods may work more effectively under non-stationary noises by learning the noises' temporal-frequency characteristics. However, most DNN methods are challenging to be implemented on mobile devices due to their heavy computation complexity. Indeed, even a few low-complexity DNN methods are proposed for real-time purposes, the robustness and the generalization degrade for different types of noise. This paper proposes a single channel DNN-based NS method for transient noise with low computation complexity. The proposed method enhanced the signal-to-noise ratio (SNR) while minimizing the speech's distortion, resulting in a superior improvement of the speech quality over different noise types, including transient noise.

Download Full-text

Speech enhancement by LSTM-based noise suppression followed by CNN-based speech restoration

EURASIP Journal on Advances in Signal Processing ◽

10.1186/s13634-020-00707-1 ◽

2020 ◽

Vol 2020 (1) ◽

Author(s):

Maximilian Strake ◽

Bruno Defraene ◽

Kristoff Fluyt ◽

Wouter Tirry ◽

Tim Fingscheidt

Keyword(s):

Speech Enhancement ◽

Noise Suppression ◽

Short Term Memory ◽

Single Channel ◽

Stationary Noise ◽

Low Snr ◽

Network Topologies ◽

Speech Restoration ◽

Lstm Network ◽

Convolutional Encoder

AbstractSingle-channel speech enhancement in highly non-stationary noise conditions is a very challenging task, especially when interfering speech is included in the noise. Deep learning-based approaches have notably improved the performance of speech enhancement algorithms under such conditions, but still introduce speech distortions if strong noise suppression shall be achieved. We propose to address this problem by using a two-stage approach, first performing noise suppression and subsequently restoring natural sounding speech, using specifically chosen neural network topologies and loss functions for each task. A mask-based long short-term memory (LSTM) network is employed for noise suppression and speech restoration is performed via spectral mapping with a convolutional encoder-decoder network (CED). The proposed method improves speech quality (PESQ) over state-of-the-art single-stage methods by about 0.1 points for unseen highly non-stationary noise types including interfering speech. Furthermore, it is able to increase intelligibility in low-SNR conditions and consistently outperforms all reference methods.

Download Full-text

Application of complex-trace analysis to seismic data for random-noise suppression and temporal resolution improvement

Geophysics ◽

10.1190/1.2196875 ◽

2006 ◽

Vol 71 (3) ◽

pp. V79-V86 ◽

Cited By ~ 19

Author(s):

Hakan Karsli ◽

Derman Dondurur ◽

Günay Çifçi

Keyword(s):

Seismic Data ◽

Trace Analysis ◽

Noise Suppression ◽

Single Channel ◽

Random Noise ◽

Synthetic Data ◽

Low Frequency ◽

Bright Spot ◽

Phase Information ◽

Resolution Improvement

Time-dependent amplitude and phase information of stacked seismic data are processed independently using complex trace analysis in order to facilitate interpretation by improving resolution and decreasing random noise. We represent seismic traces using their envelopes and instantaneous phases obtained by the Hilbert transform. The proposed method reduces the amplitudes of the low-frequency components of the envelope, while preserving the phase information. Several tests are performed in order to investigate the behavior of the present method for resolution improvement and noise suppression. Applications on both 1D and 2D synthetic data show that the method is capable of reducing the amplitudes and temporal widths of the side lobes of the input wavelets, and hence, the spectral bandwidth of the input seismic data is enhanced, resulting in an improvement in the signal-to-noise ratio. The bright-spot anomalies observed on the stacked sections become clearer because the output seismic traces have a simplified appearance allowing an easier data interpretation. We recommend applying this simple signal processing for signal enhancement prior to interpretation, especially for single channel and low-fold seismic data.

Download Full-text