python 3.x - How to create wakeword for voice speech recognition?

Question

I am working for make offline voice recognition on Raspberry pi3. I used for this job vosk-api (link:https://github.com/alphacep/vosk-api). The library works so good on Rpi3 but CPU usage is high due to continuous listen and it detect unnecessary word. Also I don't need continuous listen. So I want Rpi3 to sleep mode and just wake with wake word.

I will use Rpi3 offline. Therefore I can't use online wake word app. I tried created wake word offline and I used some libraries like Porcupine (https://github.com/Picovoice/porcupine) and rhasspy (https://github.com/rhasspy/rhasspy-wake-raven) . Porcupine was very good but I could not created own wake word. I should pay for this. Rhasppy was good for created own wake word. The downside is that it is so personal. I created my own wake word but even when I use voice recognition work it took time for the Rpi3 to recognize my voice. I think necessary so many sample for correct work.

In short, I am looking for an offline wake word create project that will generate wake words like Hello World, Raspberry wake and etc. as open source. Like the porcupine and rhasspy projects I mentioned above. I need open source project suggestions on this subject.

Any suggestions and guidance on this matter would be very pleasing to me.

Thanks

question from:https://stackoverflow.com/questions/65626026/how-to-create-wakeword-for-voice-speech-recognition

Answer 1

Waitting for answers