Voice Activity Detection (vad) Using A Single-layer Spiking Neural Network (snn) Based On The Spike Response Model (srm)

Résumé: Speech is a primary medium of human communication, and automatic speech recognition (ASR) systems aim to transcribe spoken language into text. However, noise in audio signals can significantly degrade the performance of ASR systems. Voice Activity Detection (VAD) is a crucial pre-processing step that identifies speech segments within audio signals, helping to mitigate the impact of noise. To improve the robustness of VAD systems, we propose a novel approach based on spiking neural networks (SNNs). SNNs are biologically inspired models that offer potential advantages in terms of energy efficiency and real-time processing. By leveraging the inherent capabilities of SNNs, we aim to develop a VAD system that is more resilient to noise and other adverse conditions.

Publié dans la revue: الكَلِم