Digital voice echo canceller with a TMS32020
Digital signal processing applications with the TMS320 family; Vol. 1
Self-organization and associative memory: 3rd edition
Self-organization and associative memory: 3rd edition
Neurofuzzy adaptive modelling and control
Neurofuzzy adaptive modelling and control
Speech enhancement based on neural predictive hidden Markov model
Signal Processing
Brains, Behavior and Robotics
Neurocontrol: Learning Control Systems Inspired by Neuronal Architectures and Human Problem Solving Strategies
Robust speech enhancement using amplitude spectral estimator
ICASSP '00 Proceedings of the Acoustics, Speech, and Signal Processing, 2000. on IEEE International Conference - Volume 06
IEEE Transactions on Neural Networks
A Bayesian estimation approach for speech enhancement using hiddenMarkov models
IEEE Transactions on Signal Processing
IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics
Improved MCMAC with momentum, neighborhood, and averagedtrapezoidal output
IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics
Hierarchical image coding via cerebellar model arithmetic computers
IEEE Transactions on Image Processing
Learning convergence of CMAC technique
IEEE Transactions on Neural Networks
Generalizing CMAC architecture and training
IEEE Transactions on Neural Networks
FCMAC-Yager: A Novel Yager-Inference-Scheme-Based Fuzzy CMAC
IEEE Transactions on Neural Networks
Hierarchically Clustered Adaptive Quantization CMAC and Its Learning Convergence
IEEE Transactions on Neural Networks
Hi-index | 12.05 |
Numerous attempts have been undertaken to apply the spectral subtraction method to cancel noise perturbations but these efforts have yet to produce an algorithm that is able to adapt well to the environmental changes in the perturbations. In addition, the variants of the spectral subtraction method so far proposed in the literature would require a non-voice activity detector (NVAD), for a single microphone system, to store the perturbation. This is used as an estimate for the reference signal. Inaccuracy in the perturbation estimates causes the cleaned speech to be corrupted by musical artifacts, which is unacceptable. Post processing of signals corrupted by the musical artifacts is very costly. This paper provides an alternative approach that employs associative memory for speech enhancement. Extensive comparison is made using the soft computing approaches for noise cancellation based on associative memories. A set of stereo microphones captures the corrupted speech in a vehicle and is used to point to the closest associative memory location. The Wiener filter approach is used to cancel the noise. The paper discusses novel examples of the associative memories using the cerebellum model for noise modeling. Experimental results show the potential of these novel soft computing architectures in generating and adapting the required Weiner filters to cancel perturbation even at signal to noise ratio (SNR) of less than -13dB.