Synth4Kws: Synthesized Speech for User Defined Keyword Spotting in Low Resource Environments

TL;DR

Synth4Kws demonstrates that using synthesized speech data significantly enhances custom keyword spotting models, especially in low-resource scenarios, by increasing data diversity and reducing the need for extensive real data collection.

Contribution

The paper introduces Synth4Kws, a novel framework leveraging TTS synthesized data to improve keyword spotting performance without real data, applicable across languages and keyword types.

Findings

TTS data diversity improves model performance monotonically.

Using optimal TTS data improves EER by 30.1% and AUC by 46.7%.

Mixing TTS with real data reduces real data requirements for target quality.

Abstract

One of the challenges in developing a high quality custom keyword spotting (KWS) model is the lengthy and expensive process of collecting training data covering a wide range of languages, phrases and speaking styles. We introduce Synth4Kws - a framework to leverage Text to Speech (TTS) synthesized data for custom KWS in different resource settings. With no real data, we found increasing TTS phrase diversity and utterance sampling monotonically improves model performance, as evaluated by EER and AUC metrics over 11k utterances of the speech command dataset. In low resource settings, with 50k real utterances as a baseline, we found using optimal amounts of TTS data can improve EER by 30.1% and AUC by 46.7%. Furthermore, we mix TTS data with varying amounts of real data and interpolate the real data needed to achieve various quality targets. Our experiments are based on English and single word utterances but the findings generalize to i18n languages and other keyword types.