Spectral Subtraction Filtering Research Articles

Application of Combined Filtering in Thunder Recognition

Thunder recognition is of great interest in lightning detection and physics and is widely used in short-range lightning location. However, due to the complexity of thunder, any single filtering method that is used in traditional speech noise reduction technology cannot identify well thunder from complicated background noise. In this study, the impact of four different filters on thunder recognition is compared, including low-pass filtering, least-mean-square adaptive filtering, spectral subtraction filtering, and Wiener filtering. The original acoustic signal and that filtered using different techniques are applied to a convolutional neural network, in which the thunder and background noise are classified. The results indicate that a combination of spectral subtraction and a low-pass filter performs the best in thunder recognition. The signal-to-noise ratio can be significantly improved, and the accuracy of thunder recognition (93.18%) can be improved by 3.8–18.6% after the acoustic signal is filtered using the combined filtering method. In addition, after filtering, the endpoints of a thunder signal can be better identified using the frequency domain sub-band variance algorithm.

Spectral subtraction-based filter for experimental modal analysis under harmonics excitation force

In modal analysis, measurement of input force and vibration response are crucial to accurately measure the transfer function of the structure. However, under operating condition, the force induced by operating machinery is impossible to be measured due to the sensor placement issue. In this case, the ambient response induced by the operating force should be suppressed to minimize the error in the Frequency Response Function (FRF) calculation. This paper presents the utilization of a modified spectral subtraction filter for ambient suppression. The introduction of effective ambient magnitude in gain function calculation has increased the efficiency of spectral subtraction filter. This parameter is calculated based on the phase information of the reconstructed artificial ambient response. The measurement using EMA was carried out on a motor-driven structure to verify the proposed technique. Two sets of data under shutdown and running condition were recorded to observe the effect of ambient operating force. Under the operating condition, the measured FRF show the non-identical features at operating frequencies as compared to the baseline data. The utilization of filtering process shows the ambient features contained in the transfer function was effectively suppressed. The output of filtering algorithm could provide an alternative option to perform EMA procedure under running condition.

Bark scaled oversampled WPT based speech recognition enhancement in noisy environments

The performance of speech recognition system degrades significantly in real-world environment, is a case of the acoustic mismatch between the training and operating conditions. This paper presents a two-stage approach to make a speech recognition system immune to additive and uncorrelated background noise i.e. robust. In the first stage, an oversampled wavelet packet decomposes the entire input noisy speech into seventeen nonlinear frequency subbands like the Bark scale of the human hearing system and the adaptive noise estimation based spectral subtraction filters the noisy speech from each subband signal. The oversampled WPT is linear and advantageous as it causes to overcome the shift-invariance complexity by removing the decimation after the filtering at each decomposition level. In the second stage, a nonparametric approach is used for feature extraction from filtered speech, and the parameters from the feature extraction stage are compared with the parameters extracted from speech signals stored in a template to recognize the utterance. A series of experiments are carried out to evaluate the performance of the proposed two-stage system in a variety of real environments, with and without the use of the first stage. Recognition accuracy is evaluated at the word level in a wide range of SNRs for various types of noisy environments. The experimental results show significant improvement in recognition performance at low SNR using the proposed system.

Optical and acoustic damage detection in laminated CFRP composite materials

Composite materials would gain substantial added value if it were possible to equip them with a system that could continuously monitor their damage state. In this approach, fibre-optic sensors could offer an alternative to the robust piezoelectric transducers used for acoustic emission (AE) monitoring. An intensity-modulated sensor based on the microbending concept was built and used to detect damage in laminated CFRP composite materials. Pencil lead break tests and tensile tests have been performed. The short-time Fourier Transform (STFT) has been computed and noise-reduction algorithms (adaptive filtering and spectral subtraction filtering) have also been used. The specimen final fracture can clearly be seen but transient signals were also detected before this final event. They can be correlated with the acoustic emission signals, analysed by a classical AE parameter study and with a modal acoustic emission (MAE) system. It is thus shown that the optical signal contains information on the elastic energy released whenever damage is being introduced in the host composite. Hints are that time-frequency analysis could be used to characterise this damage in a way similar to what is done for MAE.