No Word Left Behind: Mitigating Prefix Bias in Open-Vocabulary Keyword Spotting

TL;DR

This paper introduces a new benchmark and scoring method to reduce prefix bias in open-vocabulary keyword spotting, significantly improving accuracy while maintaining performance on existing datasets.

Contribution

The authors propose the Partial Overlap Benchmark and Equal-weighting Position Scoring to address prefix bias in OV-KWS, achieving substantial accuracy improvements.

Findings

EPS reduces EER from 64.4% to 29.3% on POB-Spark

Accuracy on POB-LibriPhrase improves from 87.6% to 96.8%

Adding POB data in training enhances overall benchmark performance

Abstract

Open-vocabulary keyword spotting (OV-KWS) enables personalized device control via arbitrary voice commands. Recently, researchers have explored using audio-text joint embeddings, allowing users to enroll phrases with text, and proposed techniques to disambiguate similar utterances. We find that existing OV-KWS solutions often overly bias the beginning phonemes of an enrollment, causing false triggers when negative enrollment-query-pairs share a prefix (``turn the volume up'' vs. ``turn the volume down''). We trace this to two factors: training data bias and position-biased cross-modal scoring. To address these limitations, we introduce the Partial Overlap Benchmark (POB) with two datasets, POB-Spark and POB-LibriPhrase (POB-LP), containing mismatched audio-text pairs with shared prefixes, and propose Equal-weighting Position Scoring (EPS), a lightweight decision layer. Using EPS alone reduces EER on POB-Spark from 64.4\% to 29.3\% and improves POB-LP accuracy from 87.6\% to 96.8\%, while maintaining performance on LibriPhrase and Google Speech Commands (GSC). With POB data added in training, our work achieves the best POB benchmark results while incurring the least amount of degradation on prior metrics among baselines. This degradation is most pronounced in GSC, which contains only one-word commands. We surface mitigating this trade-off as future work.