Attention-based model for detecting trigger words in streaming mode based on this paper
In this case trigger word is sheila
(la-la-la)
Download dataset
Set path to file in config
run python inference.py
and you will get probabilities of trigger word over the audio