SOTAVerified

A 510-nW Wake-Up Keyword-Spotting Chip Using Serial-FFT-Based MFCC and Binarized Depthwise Separable CNN in 28-nm CMOS

2021-01-01journal 2021Unverified0· sign in to hype

Weiwei Shan, Minhao Yang, Tao Wang, Yicheng Lu, Hao Cai, Lixuan Zhu, Jiaming Xu, Chengjun Wu, Longxing Shi, Senior Member, and Jun Yang, Member, IEEE

Unverified — Be the first to reproduce this paper.

Reproduce

Abstract

We propose a sub-µW always-ON keyword spotting (µKWS) chip for audio wake-up systems. It is mainly composed of a neural network (NN) and a feature extraction (FE) circuit. For significantly reducing the memory footprint and computational load, four techniques are used to achieve ultralow-power consumption: 1) a serial-FFT-based Mel-frequency cepstrum coefficient circuit is designed for FE, instead of the common parallel FFT. 2) A small-sized binarized depthwise separable convolutional NN (DSCNN) is designed as the classifier. 3) A framewise incremental computation technique is devised in contrast to the conventional whole-word processing. 4) Reduced computation allows a low system clock frequency, which enables near-threshold voltage operation, and low leakage memory blocks are designed to minimize the leakage power. Implemented in 28-nm CMOS technology, this µKWS consumes 0.51 µW at a 40-kHz frequency and a 0.41-V supply, with an area of 0.23 mm2. Using the Google speech command data set, 97.3% accuracy is reached for a one-word KWS task and 94.6% for a two-word task.

Tasks

Reproductions