Ok google on your own, or how to build a voice control system?

Question

Google

I want my Ok Google on Raspberry PI, or rather, an analogue, so that I can ask Malinka to turn on my favorite music from VK by voice, for example. With speech recognition, everything seems to be clear, I settled on the Google Cloud Speech API, which is convenient. However, the question arises - how to implement a command to record speech? I want something in the spirit of Ok Raspberry, and flew to recognize it. There are several problems here at once - the phrase Ok Raspberry also needs to be recognized, and since. recognition takes place on Google servers - I can’t send everything there at all that the microphone hears? This means that it is necessary to recognize this particular phrase right on the Raspberry itself. And how to organize a constant wiretapping by a microphone, before the appearance of a phrase? In general, has anyone seen implementation examples?

Answer 1

look at majordomo.smartliving.ru they play with voices and with raspberries