Boosting keyword spotting through on-device learnable user speech characteristics

TL;DR

This paper introduces a lightweight on-device learning system for keyword spotting that adapts to individual users' speech characteristics, significantly improving accuracy in TinyML applications with minimal computational resources.

Contribution

It proposes a novel architecture combining a pretrained backbone with user-aware embeddings, enabling effective on-device adaptation with low computational cost.

Findings

Error rate reduced by up to 19% on Google Speech Commands dataset.

Demonstrates effective few-shot learning in scarce data scenarios.

Requires only 23.7k parameters and 1 MFLOP per epoch for training.

Abstract

Keyword spotting systems for always-on TinyML-constrained applications require on-site tuning to boost the accuracy of offline trained classifiers when deployed in unseen inference conditions. Adapting to the speech peculiarities of target users requires many in-domain samples, often unavailable in real-world scenarios. Furthermore, current on-device learning techniques rely on computationally intensive and memory-hungry backbone update schemes, unfit for always-on, battery-powered devices. In this work, we propose a novel on-device learning architecture, composed of a pretrained backbone and a user-aware embedding learning the user's speech characteristics. The so-generated features are fused and used to classify the input utterance. For domain shifts generated by unseen speakers, we measure error rate reductions of up to 19% from 30.1% to 24.3% based on the 35-class problem of the Google Speech Commands dataset, through the inexpensive update of the user projections. We moreover demonstrate the few-shot learning capabilities of our proposed architecture in sample- and class-scarce learning conditions. With 23.7 kparameters and 1 MFLOP per epoch required for on-device training, our system is feasible for TinyML applications aimed at battery-powered microcontrollers.