Device-Directed Speech Detection: Regularization via Distillation for Weakly-Supervised Models

We address the problem of detecting speech directed to a device that does not contain a specific wake-word that is traditionally used to invoke virtual assistants (VAs). Specifically, we focus on audio that come from a touch-based invocation. Mitigating VA activation due to accidental button presses is critical for the user experience. While the majority of approaches to false trigger mitigation (FTM) are designed to detect the presence of a target keyword, inferring user intent when a keyword is not present is difficult. This also poses a challenge when creating the training/evaluation data for such systems due to inherent ambiguity in the user’s data. To this end, we propose a novel FTM approach that uses weakly-labeled training data obtained with a newly introduced data sampling strategy. While this data strategy reduces the data annotation effort, the labels of such data are noisy as the data are not annotated manually. We use these data to train an acoustics-only model for the FTM task by regularizing its loss function via knowledge distillation from an ASR-based (LatticeRNN) model. This improves the model decisions, resulting in 66% gain in accuracy, as measured by equal-error-rate (EER), over the base acoustics-only model. We also show that the ensemble of the LatticeRNN and acoustic-distilled models brings further accuracy improvement of 20%, achieving EER of 4% in the target task.

*= Equal Contributors