Abstract
Recurrent spiking neurons with lateral inhibition connection play a vital role in human’s brain functional abilities. In this paper, we propose a novel noise reduction method that is based on neuron rate coding and bio-inspired spiking neural network architecture. The excitatory-inhibitory topology in the network acts as the temporal characteristic synchrony and coincidence detector that removes uncorrelated noisy spikes. A LIF source encoder is introduced along with the network. The network uses generated binary Short-Time Fourier Transform (STFT) masks according to the rate of processed spike train, which is used to reconstruct the denoised speech signal. The technique is evaluated on noisy speech samples with 5 types of real-world additive noise with different noise strength.
Highlights
Speech enhancement methods are central to many real-world application designs e.g. hearing aids
We utilize the spiking neural networks (SNNs)’s power efficient bio-inspired computing to propose a spectrogram-based rate coding method that can contribute to the efficient lateral inhibition SNN based speech enhancement
SNN WITH LATERAL INHIBITION In [12], lateral inhibitory SNN with neighborhood connectivity [18] has been successful demonstrated on Gaussian while noise corrupted speech
Summary
Speech enhancement methods are central to many real-world application designs e.g. hearing aids. Hearing aid technologies are usually portable devices with limited size, weight, and power (SWaP) Such desirable SWaP profile increases the level of challenge of developing speech enhancement algorithms in terms of the characteristic that can satisfy both noise suppression quality and energy efficiency. It utilized multiple ARM cores and FPGAs to configure the hardware combined with PyNN [15] software API, which achieved completely scalable SNN hardware architecture with a large scale neuron capacity The emergence of these hardware technologies demonstrates a strong suitability of applying power efficient neuromorphic computing into real world mobile units. We utilize the SNN’s power efficient bio-inspired computing to propose a spectrogram-based rate coding method that can contribute to the efficient lateral inhibition SNN based speech enhancement.
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: International Journal of Machine Learning and Computing
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
