Can't get Wake Word and sound output working

Help
1

Hi Guys,

First of all thanks for the creators putting so much effort in this project.
I’m very new to Rhasspy so please bear with me. I tried to find some information in the docs and in the forum but it seems that I might miss something essential. Everything is setup according to the documentation (hopefully :thinking:).

Here is my setup:
Master

  • Debian 9 in a VM.
  • running Rhasspy 2.5.0
  • Docker Setup
  • German Profile

Satellite

I was able to create some sentences and also test them using the Wake Button from the GUI. It’s working on both the Satellite and the Master. However I’m a bit lost configuring the correct sound output on the satellite . I have a 3.5mm headset connected directly to the MicHat2 and can’t hear anything. I would expect to get some audio feedback when pressing the wake button or at least some audio output when using the text to speak feature. But there is nothing. Also I can’t hear anything using the play recording button so I can’t say how good the audio input is.
On the other hand the Wake Word detection doesn’t seem to respond. I tried several Wake Words without any luck. I first used the German profile but read here in the forum to better use English profile for the satellite. Unfortunately that didn’t help.
Where do I find the logfiles so I can provide them when needed?

Thanks in advance.
Best regards Marc.

2

regarding the audio issue:
Did you select the correct audio output ?
Does the audio output on the pi zero w work with other applications? (like if you just use aplay with a audio file from the commandline)
in your config CARD=headphone is selected and atleast for me the 2mic hat is called CARD=seeed2micvoicec .

regarding the issue with the wakeword:
Did you already try setting the sensitivity for snowboy to something like 0.8 or higher?

1 Like
3

Thanks @moqart for your response.

I’ve tried several but probably not the correct one. Now it works. :+1:

yes - I used arecord -f cd -Dhw:1 | aplay -Dhw:1 command according to respeaker docs and got correct readback.

I noticed that after changing the (audio?) settings i had to restart the container - otherwise I got some timeouts when using e.g. tts. I’ll further check that tomorrow.

I also changed some settings for snowboy including the sensitivity and now it seems to work. :thinking: :+1: :+1: :+1: