respeaker_python_library
respeaker_python_library copied to clipboard
Reduce false positives
Any idea on how to reduce the amount of false positives I get on a custom wake word?
The system does well recognizing the words when I speak them, however, it also constantly misinterprets environment sounds as the wake word. Sometimes, it's just noise that sets it off, like shuffling feet at random points of the day.
I know PocketPhoenix is one of the least accurate systems, but what can I do to make it less sensitive to environment sounds and more accurate with wake word?