Complementary Language Model and Parallel Bi-LRNN for False Trigger Mitigation

Abstract: False triggers in voice assistants are unintended invocations of the assistant, which not only degrade the user experience but may also compromise privacy. False trigger mitigation (FTM) is a process to detect the false trigger events and respond appropriately to the user. In this paper, we propose a novel solution to the FTM problem by introducing a parallel ASR decoding process with a special LLM trained from "out-of-domain" data sources. Such LLM is complementary to the existing LLM optimized for the assistant task. A bidirectional lattice RNN (Bi-LRNN) classifier trained from the lattices generated by the complementary LLM shows a $38.34\%$ relative reduction of the false trigger (FT) rate at the fixed rate of $0.4\%$ false suppression (FS) of correct invocations, compared to the current Bi-LRNN model. In addition, we propose to train a parallel Bi-LRNN model based on the decoding lattices from both LLMs, and examine various ways of implementation. The resulting model leads to further reduction in the false trigger rate by $10.8\%$.