Towards noise robust trigger-word detection with contrastive learning\n pre-task for fast on-boarding of new trigger-words

Trigger-word detection plays an important role as the entry point of user's\ncommunication with voice assistants. But supporting a particular word as a\ntrigger-word involves huge amount of data collection, augmentation and\nlabelling for that word. This makes supporting new trigger-words a tedious and\ntime consuming process. To combat this, we explore the use of contrastive\nlearning as a pre-training task that helps the detection model to generalize to\ndifferent words and noise conditions. We explore supervised contrastive\ntechniques and also propose a novel self-supervised training technique using\nchunked words from long sentence audios. We show that both supervised and the\nnew self-supervised contrastive pre-training techniques have comparable results\nto a traditional classification pre-training on new trigger words with less\ndata availability.\n