I installed Rhasspy via the HA addon and it all appears to be working.
This HA install is on a VM on a Qnap Nas.
my main pc where i usually use most things is a Windows pc
Is there any way to send voice commands via my windows pc to this Rhasspy install? the Qnap is not in an ideal place to connect a mic and speakers to it. and doesnt have a screen so i need to be at a pc when im doing the training etc.
Is there a way to send voice from Windows?
easiest way, but not good for permanent usage is to just use the web interface. it has an option to upload wav files.
a better way would probably be to install rhasspy locally on your pc and use it as a satellite. seeing as you mentioned windows, I would suggest installing a vm like virtualbox or vmware player (free for a single virtual machine) with a linux on it. if you are really good with software, you also could try wsl, but wsl 1 never really ran for me and wsl 2 can’t pass through sound and devices without a huge struggle (installing pulseaudio on windows and piping audio via network was the latest I know about for sound) so it would not be advisable.
but even if you have a working rhasspy satellite on your windows pc, this is not a good solution for a voice assistant. it would only work when your pc is on and the vm is running. if you just want to train on it and later connect a mic and speakers to your nas, you could also run into trouble if you train a custom wake word if it is a different mic because the sound will be different.
if you want a permanent voice assistant, then I would suggest investing in some hardware to be a satellite, either a pi with a mic hat, or an esp with a build in or self built mic array. there are quite a few topics about hardware for this kind of thing. if you want it all to be done by one device, then the best bet would be a pi4, since that runs rhasspy really well. only hardware you should not get is a pi zero, because while they can run rhasspy as a satellite, there are quite a few esps that cost about as much that run better if the task is just stream voice to the base station and play audio sent by the base station and as far as I have read, that is the only way to really use the pi zero with rhasspy, and it comes with the “bloat” of a full rhasspy install that you can’t use because the hardware can’t handle it in real time.
also, there is a mobile app that works as a satellite, not sure about the state of development but last I heard it was working pretty well. somewhere in this forum I also read about a project that uses a browser as an input for a satellite, but I am not sure if that ever got far enough to be usable.
Hi, thanks for the in depth reply.
I was kind of looking for a quick simple way to do the initial learning curve but its a really good point that the mic will change what the audio will sound like so the training really needs to be done on the actual satellite hardware.
I did just get a couple of respeaker hats but i did go with a pi zero W not realising rhasppy wouldnt run well on it.
ill grab a couple of 3b+ to make my satellites.
im sure ill find a use for the pi w if i can put it somewhere i dont immediatelly lose it like last time.
Ill also try the mobile app. it seems to have a couple of pull requests open and no activity but installs and connects so should work i am hoping.