Progressive Voice Trigger Detection: Accuracy vs Latency

TL;DR

This paper introduces a progressive voice trigger detection system that balances accuracy and latency by using additional audio context after trigger phrases, significantly reducing false rejections with minimal delay.

Contribution

The work proposes a two-stage architecture that adaptively delays decisions to improve trigger detection accuracy without substantially increasing latency.

Findings

66% relative reduction in false rejection rate

Only 3% of triggers delayed for additional context

Negligible increase in overall latency

Abstract

We present an architecture for voice trigger detection for virtual assistants. The main idea in this work is to exploit information in words that immediately follow the trigger phrase. We first demonstrate that by including more audio context after a detected trigger phrase, we can indeed get a more accurate decision. However, waiting to listen to more audio each time incurs a latency increase. Progressive Voice Trigger Detection allows us to trade-off latency and accuracy by accepting clear trigger candidates quickly, but waiting for more context to decide whether to accept more marginal examples. Using a two-stage architecture, we show that by delaying the decision for just 3% of detected true triggers in the test set, we are able to obtain a relative improvement of 66% in false rejection rate, while incurring only a negligible increase in latency.