Speech signal enhancement through adaptive wavelet thresholding
Speech Communication
Robust voice activity detection using perceptual wavelet-packet transform and Teager energy operator
Pattern Recognition Letters
Speech Enhancement, Gain, and Noise Spectrum Adaptation Using Approximate Bayesian Estimation
IEEE Transactions on Audio, Speech, and Language Processing
Blind Spatial Subtraction Array for Speech Enhancement in Noisy Environment
IEEE Transactions on Audio, Speech, and Language Processing
IEEE Transactions on Audio, Speech, and Language Processing
IEEE Transactions on Audio, Speech, and Language Processing
A Soft Voice Activity Detection Using GARCH Filter and Variance Gamma Distribution
IEEE Transactions on Audio, Speech, and Language Processing
Simultaneous Detection and Estimation Approach for Speech Enhancement
IEEE Transactions on Audio, Speech, and Language Processing
Robust Speech Dereverberation Using Multichannel Blind Deconvolution With Spectral Subtraction
IEEE Transactions on Audio, Speech, and Language Processing
IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics
Hi-index | 0.00 |
In this paper an adaptive fuzzy filter, based on fuzzy system, is proposed for speech signal enhancement and automatic speech recognition accuracy. In the past two decades the basic wavelet thresholding-algorithm has been widely studied and is common applied to filter noise. In the proposed system adaptive wavelet thresholds are generated and controlled by fuzzy rules concerning the presence of speech in noise. First an amplified voice activity detector is designed to improve performance on SNR lower than 5dB. Then an adaptive threshold decision module based on fuzzy inference system is proposed. In this fuzzy inference system overall relations between speech and noise are summarized into seven fuzzy rules and four linguistic variables, which are used to detect the state of signals. The adaptive threshold and membership functions are optimally obtained by particle swarm optimization algorithm so the SNR of the filter output for training signal data can be maximized. Experimental results reveal that our proposed system effectively increases the SNR and the recognition rate of speech.