LiCo-Net: Linearized Convolution Network for Hardware-efficient Keyword Spotting

TL;DR

LiCo-Net is a hardware-efficient keyword spotting architecture optimized for low-power microcontrollers, combining int8 linear operators and streaming convolutions to improve efficiency while maintaining detection accuracy.

Contribution

The paper introduces LiCo-Net, a dual-phase system that enhances hardware efficiency for keyword spotting by integrating int8 linear operators during inference and streaming convolutions during training.

Findings

LiCo-Net reduces cycles by 40% on HiFi4 DSP.

LiCo-Net outperforms SVDF in hardware efficiency.

Maintains high detection performance with lower computational cost.

Abstract

This paper proposes a hardware-efficient architecture, Linearized Convolution Network (LiCo-Net) for keyword spotting. It is optimized specifically for low-power processor units like microcontrollers. ML operators exhibit heterogeneous efficiency profiles on power-efficient hardware. Given the exact theoretical computation cost, int8 operators are more computation-effective than float operators, and linear layers are often more efficient than other layers. The proposed LiCo-Net is a dual-phase system that uses the efficient int8 linear operators at the inference phase and applies streaming convolutions at the training phase to maintain a high model capacity. The experimental results show that LiCo-Net outperforms single-value decomposition filter (SVDF) on hardware efficiency with on-par detection performance. Compared to SVDF, LiCo-Net reduces cycles by 40% on HiFi4 DSP.