Abstract:Signal denoising is a key preprocessing step for many applications, as the performance of a learning task is closely related to the quality of the input data. In this paper, we apply a signal processing based deep neural network architecture, a learnable extension of the wavelet packet transform. As main advantages, this model has few parameters, an intuitive initialization and strong learning capabilities. Moreover, we show that it is possible to easily modify the parameters of the model after the training step to tailor to different noise intensities. Two case studies are conducted to compare this model with the state of the art and commonly used denoising procedures. The first experiment uses standard signals to study denoising properties of the algorithms. The second experiment is a real application with the objective to remove audio background noises. We show that the learnable wavelet packet transform has the learning capabilities of deep learning methods while maintaining the robustness of standard signal processing approaches. More specifically, we demonstrate that our approach maintains excellent denoising performances on signal classes separate from those used during the training step. Moreover, the learnable wavelet packet transform was found to be robust when different noise intensities, noise varieties and artifacts are considered.

The Construction of Wavelet Network for Speech Signal Processing

Wavelet Network with OLS Optimization for Speech Signal Processing

Denoising Speech Based on Deep Learning and Wavelet Decomposition

A Novel Structure of Wavelet Neural Network and Its Learning Algorithm

Distributed wavelet neural networks

On The Construction Of Support Wavelet Network

Construction of a New Adaptive Wavelet Network and Its Learning Algorithm

Near-optimal Construction of Wavelet Networks for Nonlinear System Modeling

Orthogonal Wavelet Networks for Nonlinear Dynamic System Identification

Nonlinear System Identification Using Orthogonal Wavelet Networks

Application of wavelet transform and neural network to the classification of underwater wideband echo

Improved Speech Enhancement with the Wave-U-Net

Complex-Valued Wavelet Network

Wavelet J-Net: A Frequency Perspective on Convolutional Neural Networks

Modeling and Optimization of Microwave Circuits and Devices Using Wavelet Neural Networks

A Hierarchical Evolutionary Algorithm for Constructing and Training Wavelet Networks.

Novel Algorithm of Wavelet Network Structural Design

Wavelet neural network blind equalization with cascade filter base on RLS in underwater acoustic communication.

Speech preprocessing and enhancement based on joint time domain and time-frequency domain analysis

Robust Time Series Denoising with Learnable Wavelet Packet Transform

Wavelet Neural Networks for Adaptive Equalization by Using the Orthogonal Least Square Algorithm