Acoustic Scene Classification Technique for Active Noise Control

A survey on active noise control in the past decade–Part II: Nonlinear systems

Hybrid noise control is the combination of both active noise and passive noise control. For high-frequency noise, we generally use passive noise control technique, and for low-frequency noise of less than 200 Hz, we use active noise control technique. Passive noise control is achieved by introducing passive techniques such as enclosures, physical barriers, and silencers to attenuate the undesired noise, whereas active noise control (ANC) is achieved by introducing a cancelling “anti-noise” signal generated by a processor or electronics circuit. Digital filters can be used for this purpose with an appropriate adaptive algorithms such as FXLMS, FXNLMS, FBLMS, SE-FXLMS, and FULMS. The development of an ANC simulation model can be divided into two parts: the modelling of primary noise and the modelling of secondary or counter noise. The anti-noise generated by the ANC system can be modelled only if we have the information of primary noise. For successful simulation of secondary noise, we have to model the control system and its interaction with the surrounding acoustical space. This article describes the practical application of active noise control as well as passive noise control to a server room, greatly reducing the level of noise in the whole area through appropriate adaptive algorithms. The results obtained using different algorithms are compared and analyzed.

In Acoustic Scene Classification (ASC) two major approaches have been followed. While one utilizes engineered features such as mel-frequency-cepstral-coefficients (MFCCs), the other uses learned features that are the outcome of an optimization algorithm. I-vectors are the result of a modeling technique that usually takes engineered features as input. It has been shown that standard MFCCs extracted from monaural audio signals lead to i-vectors that exhibit poor performance, especially on indoor acoustic scenes. At the same time, Convolutional Neural Networks (CNNs) are well known for their ability to learn features by optimizing their filters. They have been applied on ASC and have shown promising results. In this paper, we first propose a novel multi-channel i-vector extraction and scoring scheme for ASC, improving their performance on indoor and outdoor scenes. Second, we propose a CNN architecture that achieves promising ASC results. Further, we show that i-vectors and CNNs capture complementary information from acoustic scenes. Finally, we propose a hybrid system for ASC using multi-channel i-vectors and CNNs by utilizing a score fusion technique. Using our method, we participated in the ASC task of the DCASE-2016 challenge. Our hybrid approach achieved 1st rank among 49 submissions, substantially improving the previous state of the art.

Since the inception of Neonatal Intensive Care Units (NICU) many improvements have been done in them time to time to make them safer and better for the infants. DSP processors played an important role in efficient implementation of algorithms, which helped in reducing the noises generated by medical equipment and other external factors. A zone of silence (ZoS) is created in the vicinity of baby's head using different Active Noise Control (ANC) techniques. The focus was on designing an active noise control system, which not only eliminates the noise but also plays the good sounds for the baby along with other tactics that can be helpful in monitoring without disturbing the baby. In this paper various active noise control techniques are reviewed which are developed over the period of time for NICUs. The evolution of new techniques takes place to overcome the earlier problems and pitfalls in designs and implementation, cost effective systems, and different algorithms and techniques used to enhance the quality of noise control.

With the advancement of digital signal technology, active noise control technique has gradually received more attention for the low-frequency noise in the vehicle. The active noise control apparatus and method have shortcomings in the theory and algorithm. Both acoustic devices researched on meeting the application requirements and adaptive active noise control algorithms are effective ways to improve the practicability of active noise control technique. The paper reviews various vehicle interior active noise control apparatuses in recent years. It has been found that the vehicle interior active noise control apparatuses can be classified as the extracting signal sources and the control strategy between cancelling signal and feedback signal. From review of the patents presented, it is found that there are many patents covering the active noise control method inside vehicle. However, there is a limitation in controlling speed and precision. Therefore, it is recommended that the development should be focused on the design of the control section of vehicle interior noise, which will improve the sound quality and save the space. Moreover, the cost of the new apparatus will be reduced by introducing intelligent controller. Keywords: Acoustic devices, active noise control strategy, algorithm, cancelling signal, feedback signal, vehicle interior noise.

Acoustic scene classification is the task of identifying the scene from which the audio signal is recorded. Convolutional neural network (CNN) models are widely adopted with proven successes in acoustic scene classification. However, there is little insight on how an audio scene is perceived in CNN, as what have been demonstrated in image recognition research. In the present study, the Class Activation Mapping (CAM) is utilized to analyze how the log-magnitude Mel-scale filter-bank (log-Mel) features of different acoustic scenes are learned in a CNN classifier. It is noted that distinct high-energy time-frequency components of audio signals generally do not correspond to strong activation on CAM, while the background sound texture are well learned in CNN. In order to make the sound texture more salient, we propose to apply the Difference of Gaussian (DoG) and Sobel operator to process the log-Mel features and enhance edge information of the time-frequency image. Experimental results on the DCASE 2017 ASC challenge show that using edge enhanced log-Mel images as input feature of CNN significantly improves the performance of audio scene classification.

Acoustic scene classification and related tasks have been dominated by Convolutional Neural Networks (CNNs). Top-performing CNNs use mainly audio spectograms as input and borrow their architectural design primarily from computer vision. A recent study has shown that restricting the receptive field (RF) of CNNs in appropriate ways is crucial for their performance, robustness and generalization in audio tasks. One side effect of restricting the RF of CNNs is that more frequency information is lost. In this paper, we perform a systematic investigation of different RF configuration for various CNN architectures on the DCASE 2019 Task 1.A dataset. Second, we introduce Frequency Aware CNNs to compensate for the lack of frequency information caused by the restricted RF, and experimentally determine if and in what RF ranges they yield additional improvement. The result of these investigations are several well-performing submissions to different tasks in the DCASE 2019 Challenge.

It is difficult to reduce the noise of a moving evaluation point with Active Noise Control (ANC) technique. This is because that ANC technique can control only narrow area. Therefore to expand quiet area is necessary. Although we showed that it is possible to expand quiet area against pure tone in the previous study, but it is not yet capable against random noise. In this paper, we propose the new method for expanding quiet area against random noise. Two conditions, noise reduction and noise passage conditions, for an error microphone are investigated. Under the noise reduction condition, noise is reduced at the position of error microphone. On the other hand, under noise passage condition, all control sound are cancelled each other at the position of error microphone. These conditions depend on a standard frequency. When the standard frequency is same to the noise frequency of pure tone, it makes noisier area narrow to set error microphones meeting these conditions. Accordingly, we think that to meet these conditions at many standard frequencies make noisier area narrow in random noise effectively. The control effect is shown by the numerical simulations.

It is important to remove large environmental noise in measurement using SQUIDs without magnetic shielding. Here, the environmental noise has been reduced by an active noise control (ANC) system in the MHz radio-frequency region. The ANC system including a phase and amplitude control circuit has been designed to allow precise adjustments of the phase and the amplitude of the anti-phase wave against the noise waves in the MHz region. In this paper, a sinusoidal wave in the MHz range was used as the radio frequency interference. When a SQUID was used for a signal detector, the ANC system suppressed this sinusoidal wave to the white noise level by about 54 dB. The performance of the ANC system for the chemical detection by nuclear quadrupole resonance (NQR) was checked in a simulated environment. One was the case of a real AM radio broadcasting signal with the dummy NQR signal and another was the real NQR signal with a sinusoidal wave as noise. For both cases, the ANC showed good performance for MHz range noise cancellation.

In practical active noise control (ANC) systems, the primary path and the secondary path may be nonlinear and time-varying. It has been reported that the linear techniques used to control such ANC systems exhibit degradation in performance. In addition, the actuators of an ANC system very often have nonminimum-phase response. A linear controller under such situations yields poor performance. A novel functional link artificial neural network (FLANN)-based simultaneous perturbation stochastic approximation (SPSA) algorithm, which functions as a nonlinear mode-free (MF) controller, is proposed in this paper. Computer simulations have been carried out to demonstrate that the proposed algorithm outperforms the standard filtered-x least mean square (FXLMS) algorithm, and performs better than the recently proposed filtered-s least mean square (FSLMS) algorithm when the secondary path is time-varying. This observation implies that the SPSA-based MF controller can eliminate the need of the modeling of the secondary path for the ANC system.

Based on the theoretical analysis of literature, saturation in measured signal of active noise control (ANC) systems will degrade the convergence speed. However, the experiments show that the saturated input signal can speed up the convergence of the narrow-band ANC systems. This paper intends to remodel the saturation effects for feedforward and feedback ANC systems. Combining the action of analog-to-digital converters (ADC), the mathematical expression and block diagrams are proposed to model the saturation effects in the practical ANC systems. The derivation and simulation results show that since the saturation is able to amplify the principle component of signal, the convergence would be speeded up.

Acoustic scene classification (ASC) is a problem related to the field of machine listening whose objective is to classify/tag an audio clip in a predefined label describing a scene location (e. g. park, airport, etc.). Many state-of-the-art solutions to ASC incorporate data augmentation techniques and model ensembles. However, considerable improvements can also be achieved only by modifying the architecture of convolutional neural networks (CNNs). In this work we propose two novel squeeze-excitation blocks to improve the accuracy of a CNN-based ASC framework based on residual learning. The main idea of squeeze-excitation blocks is to learn spatial and channel-wise feature maps independently instead of jointly as standard CNNs do. This is usually achieved by some global grouping operators, linear operators and a final calibration between the input of the block and its obtained relationships. The behavior of the block that implements such operators and, therefore, the entire neural network, can be modified depending on the input to the block, the established residual configurations and the selected non-linear activations. The analysis has been carried out using the TAU Urban Acoustic Scenes 2019 dataset (https://zenodo.org/record/2589280) presented in the 2019 edition of the DCASE challenge. All configurations discussed in this document exceed the performance of the baseline proposed by the DCASE organization by 13\% percentage points. In turn, the novel configurations proposed in this paper outperform the residual configurations proposed in previous works.

Active noise control techniques have been studied in recent years as an available method of machinery noise reduction. Especially, there are many applications of them to ductlike structures where sound waved radiated from noise sources can be assumed to be traveling as plane waves. In most of these cases, the lengths by ducts were assumed to be infinitely long, and acoustic properties such as reflections and resonances inside the ducts have seldom been considered. In actual machines, however, the duct lengths ate often short because of the limit of machine dimensions; hence a basic study on active noise control for such short-length ducts is needed. This paper describes the modeling method of an active noise control system using a fixed FIR filter for short-length ducts. The results of simulation bclculations were compared with experimental data and good agreement was obtained.

Frequency-based CNN and attention module for acoustic scene classification

Hierarchical classification for acoustic scenes using deep learning