I’m on rhasspy 2.4.20, running from vevn, using the run-venv.sh script on a Raspberry PI 4, using a playstation eye as the mic and the regular line out from the Raspberry as the sound output.
When I invoke the system, nothing happens, as if the wake function is not working. If I try to use the Web UI (hold to record, choosing the onboard mic) it correctly recognizes some sentences. But If I just try “Hey Rhasspy” , or anything at all I can’t see anything going on.
Is there a log file somewhere? I see that log messages scroll by, but are they also on some file?
What would be a good hint to hunt for why it is not working?
In the web interface on the right there is a tab called ‘Log’. When you say ‘Hey porcupine’ any output will appear here. Also if the word is recognized the ‘R’ icon on the top left should start to spin.
I don’t think it will respond to ‘Hey Rhasspy’, look at what Wake Word is set, see the ‘Settings’ tab.
I’ve finally managed to get something working…I had to change the settings to Mycroft Precise and then download some pre-trained models.
Now I see that it sometimes recognizes my command. Sometimes it takes a long time to be able to recognize again. I’m still trying to understand what is going on after it wakes up (and maybe timeouts later?).
Is there a way to debug the whole process? Maybe increase the log to a higher verbose setting?
I get 1 out of five replies from the wake word part and less than 1 out of 5 replies after the wake word. Most of the time, Rhasspy stays totally unresponsive for a long time (maybe wating for a timeout somewhere?).