You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I had updated to a version 1.0 voice-controller add-on on a 0.8.1 gateway that had the previous voice-controller version. When I plugged the speaker/mic back into it (had been detached), I disabled/enabled the voice-controller. Note to self -- because packages are uninstalled and re-installed for this process, you have to go have coffee or something while waiting (takes a long time on a little RPi.) After updating the config to "hey snips" and waiting some more, then the wake word was being detected, but it failed on a simple command, "turn on the balcony" (a light). I suspect the problem is that after reconfig/retrain/reinstall/restart, somehow the controller didn't "start up" with a known set of thing names in its language model (LM). After I had deleted a thing from my UI, I saw that the voice-controller had done a "retrain" event (saw in log). When I tried a voice command again, it recognized all my thing names. :) So I think that after every "configure" or "enable" of the voice-controller, we should make sure to also run the "retrain" process to make sure the LM is up to date.
The text was updated successfully, but these errors were encountered:
I had updated to a version 1.0 voice-controller add-on on a 0.8.1 gateway that had the previous voice-controller version. When I plugged the speaker/mic back into it (had been detached), I disabled/enabled the voice-controller. Note to self -- because packages are uninstalled and re-installed for this process, you have to go have coffee or something while waiting (takes a long time on a little RPi.) After updating the config to "hey snips" and waiting some more, then the wake word was being detected, but it failed on a simple command, "turn on the balcony" (a light). I suspect the problem is that after reconfig/retrain/reinstall/restart, somehow the controller didn't "start up" with a known set of thing names in its language model (LM). After I had deleted a thing from my UI, I saw that the voice-controller had done a "retrain" event (saw in log). When I tried a voice command again, it recognized all my thing names. :) So I think that after every "configure" or "enable" of the voice-controller, we should make sure to also run the "retrain" process to make sure the LM is up to date.
The text was updated successfully, but these errors were encountered: