Spike-based encoding and learning of spectrum features for robust sound recognition

Biological evidence suggests that local time-frequency (LTF) information can be utilized to improve the recognition rate of sounds in the presence of noise. However, most of conventional methods use stationary (frequency-based) features which are not robust to noise, as each stationary feature contains a mixture of spectral information from both noise and signal. This paper proposes a spike-timing based model to encode and learn the LTF features extracted from sound spectrogram using spiking neural networks (SNNs), named LTF-SNN. In this model, we encode the reliable LTF features into spike train patterns and train with different spike-based learning rules. We analyze the efficacy of the spike-based feature encoding method and the recognition performance of the model by using two classes of SNN learning algorithms: ReSuMe and Tempotron. Utilizing the temporal coding and learning, networks of spiking neurons can effectively perform robust sound recognition tasks. Experimental results demonstrate that the model achieves superior performance in mismatched conditions compared with benchmark approaches.