A 23 $\mu$W Keyword Spotting IC with Ring-Oscillator-Based Time-Domain Feature Extraction

TL;DR

This paper introduces a low-power, fully time-domain keyword spotting integrated circuit that leverages ring-oscillator-based feature extraction, achieving high accuracy and low latency suitable for edge applications.

Contribution

It presents the first KWS IC using ring-oscillator-based time-domain feature extraction, demonstrating improved scalability and power efficiency in a 65 nm CMOS process.

Findings

Achieves 86% accuracy on Google Speech Command Dataset

Consumes only 23 μW power in total

Operates with 12.4 ms latency

Abstract

This article presents the first keyword spotting (KWS) IC which uses a ring-oscillator-based time-domain processing technique for its analog feature extractor (FEx). Its extensive usage of time-encoding schemes allows the analog audio signal to be processed in a fully time-domain manner except for the voltage-to-time conversion stage of the analog front-end. Benefiting from fundamental building blocks based on digital logic gates, it offers a better technology scalability compared to conventional voltage-domain designs. Fabricated in a 65 nm CMOS process, the prototyped KWS IC occupies 2.03mm$^{2}$ and dissipates 23 $\mu$W power consumption including analog FEx and digital neural network classifier. The 16-channel time-domain FEx achieves 54.89 dB dynamic range for 16 ms frame shift size while consuming 9.3 $\mu$W. The measurement result verifies that the proposed IC performs a 12-class KWS task on the Google Speech Command Dataset (GSCD) with >86% accuracy and 12.4 ms latency.