I recently installed Rhasspy and am trying to get it to work but I’ve run into an unexpected issue. I set everything to the default recommended and when I say “Porcupine” it chimes and shows that it listens for about 10 seconds. But no speech is ever recognized and when I play the WAV file it is ~0.25 seconds long of silence. I know the microphone is working as it wakes up to the wake word and in statistics I can see audio statistics I can see the energy line up to my speech or hand claps. So I’m not sure how this is happening or how to fix this. I installed Rhasspy as an add on to home assistant which is running as a VM.
That is odd.
- You can hear the chime, which suggests your speaker is working correctly.
- Detecting your wake word suggests the microphone is working.
Could you post here your Rhasspy configuration. It’s a pain to post a lot of printscreens, but the “Advanced” option in the “Settings” menu gives a nice text summary which is easy to copy and paste here. Remember to blank out any passwords.
I tried restarting multiple times but didn’t get anywhere. I ended up loading Rhasspy as a docker image instead and have it up and running that way. Unfortunately this means I will need two speakers as home assistant has one and Rhasspy docker will need a separate one. My next task will be to try to find a way to connect them both to a remote media server or something if that’s possible. But for now it is working.
If you’re still interested in the config I’ll still post it below. Thanks for the help!
{
“dialogue”: {
“system”: “rhasspy”
},
“handle”: {
“system”: “hass”
},
“home_assistant”: {
“access_token”: “REDACTED”
},
“intent”: {
“system”: “fsticuffs”
},
“microphone”: {
“system”: “pyaudio”
},
“sounds”: {
“aplay”: {
“volume”: “0.2”
},
“system”: “aplay”
},
“speech_to_text”: {
“system”: “kaldi”
},
“text_to_speech”: {
“nanotts”: {
“language”: “en-GB”,
“volume”: “0.1”
},
“system”: “nanotts”
},
“wake”: {
“porcupine”: {
“keyword_path”: “porcupine_linux.ppn”
},
“system”: “porcupine”
}
}
Huh ? why do you need 2 speakers ? Is the HA one for playing music, and the Rhasspy one for text-to-speech ?