问题
I am using the Google python speech recognition on a Raspberry Pi and it works really well. However, it is constantly listening and I would prefer to have a wake word so that it only listens when I need it to.
At the moment it listens until there is a silence (or a return to whatever the ambient background noise is) and then sends the sound clip to the API for speech to text recognition. Is there a way for me to shorten the length of time that it is listening? This way it can be listening to only one word at a time and waiting for whatever wake word I have set.
回答1:
It's better to implement the wake-word detection offline. Have a look at Snowboy. They let you upload a few recordings of your wake word and you can get a trained model that you can then include in your application.
来源：https://stackoverflow.com/questions/44639836/google-python-speech-recognition-with-wake-word