Wake word support

[timmo001]

Looking to implement openwakeword, this would probably require converting the application to use the wyoming protocol

[timmo001]

Looking for experts on this subject, any help would be great

[timmo001]

This might be an example of what can be used:

https://github.com/dscripka/openWakeWord/blob/main/examples/web/README.md

[timmo001]

This example contains an example streaming to a python server with openwakeword:

For custom models, I have only been able to get this working not on windows (wsl works also), due to the requirement for tensorflow-runtime:

aidan@XPS-15-9530:~/openWakeWord/examples/web$ python3 streaming_server.py --model_path /home/aidan/computer_v2.tflite

[timmo001]

I think the end goal we want to accomplish is to stream audio to the wyoming-openwakeword / home assistant server, and use the selected pipeline's wake word, instead of implementing our own. While running locally is nice, it doesnt give us the flexibility to use whatever wake word the user wants.

It also would be a pain to make the above cross platform, Windows (non-wsl) being the problem this time