ED-sKWS: Early-Decision Spiking Neural Networks for Rapid,and Energy-Efficient Keyword Spotting

TL;DR

This paper introduces ED-sKWS, an energy-efficient spiking neural network with early decision capability for rapid keyword spotting, reducing latency and energy use while maintaining accuracy.

Contribution

The paper proposes ED-sKWS, a novel SNN model with an early-decision mechanism and a new loss function, improving speed and energy efficiency in keyword spotting tasks.

Findings

Maintains 61% accuracy timesteps with 52% energy consumption.

Outperforms non-early-decision SNN models in speed and energy efficiency.

Introduces the SC-100 dataset for early-decision KWS evaluation.

Abstract

Keyword Spotting (KWS) is essential in edge computing requiring rapid and energy-efficient responses. Spiking Neural Networks (SNNs) are well-suited for KWS for their efficiency and temporal capacity for speech. To further reduce the latency and energy consumption, this study introduces ED-sKWS, an SNN-based KWS model with an early-decision mechanism that can stop speech processing and output the result before the end of speech utterance. Furthermore, we introduce a Cumulative Temporal (CT) loss that can enhance prediction accuracy at both the intermediate and final timesteps. To evaluate early-decision performance, we present the SC-100 dataset including 100 speech commands with beginning and end timestamp annotation. Experiments on the Google Speech Commands v2 and our SC-100 datasets show that ED-sKWS maintains competitive accuracy with 61% timesteps and 52% energy consumption compared to SNN models without early-decision mechanism, ensuring rapid response and energy efficiency.