I adjusted it only slightly by allocating 4GiB and then 6GiB to the system during install.
CPU being used by my Proxmox hardware is an Intel i7 4790K. Old but not ancient
After installing both Piper and Whisper, I am seeing Piper fail to start, and seeing this error in logs:
RuntimeError: NumPy was built with baseline optimizations:
(X86_V2) but your machine doesn't support:
(X86_V2).
[14:51:19] INFO: Service exited with code 1 (by signal 0)
Searching seems to point me in two directions - either the CPU is too old and doesn’t support it, or I have an old version of NumPy installed by the community-script installer for Proxmox and might just need to update it?..
Is anyone able help confirm what my next steps should be? I can’t seem to find a command that works through Proxmox console to output my NumPy version either, both ‘print’ and ‘import’ commands are not working
Piper and Whisper should be HA Apps and install easily.
Is that script you are running on proxmox installing HAOS?
If it’s installing an old deprecated Core or Supervised version, that is your problem.
What versions are you running - what does this say?
Correct, I used that community script to install HA as a VM on Proxmox. It’s easier than doing it manually. It shouldn’t be a dated thing…
The Home Assistant team provides an official KVM image for this installation, making it a supported and recommended method. While considered an easy/medium difficulty installation, basic familiarity with Linux is recommended.
The following script automates the process of creating a Virtual Machine (VM) using the official KVM (qcow2) disk image provided by the Home Assistant Team. It involves finding, downloading, and extracting the image, defining user-defined settings, importing and attaching the disk, setting the boot order, and starting the VM. It supports various storage types, and does not involve any hidden installations.
I used the VM instead of LXC (container) since the container does not support addons
Specific doco here Proxmox VE Scripts
My HA version info currently:
Installation method Home Assistant OS
Core 2026.7.4
Supervisor 2026.07.3
Operating System 18.1
Frontend 20260624.6
It installed and when I browse the UI to Settings > Apps it says ‘Running’
The logs seem to be repeating this message like it’s in a loop of initializing and then killing the process and trying again.
The reason I was looking at logs is because I was following this guide to set up local voice assist
When I got to the step of selecting the TTS and STT neither Piper or Whisper were available in the drop-down menus. So naturally I looked at the log, starting with Piper, and see what appears to be this loop.
I’ve already tried removing and reinstalling with no success too.
Maybe this app reqiure AVX512 ISA ?
And installing NumPy where ? You have done that not in the container i suppose.
I didn’t install anything manually outside of what the script installer for Proxmox did.
Can you elaborate on the app requirement? I would be surprised if this i7 didn’t have something Piper needs since the HA documentation says you can even run this local-only voice assist on a Raspberry Pi which would be way more ancient in that regard
Thanks! I was apprehensive to change anything from the deployment script, but after doing some reading about the differences between the deployed qemu and host settings I went ahead and made the change so the HA LXC can use all the CPU features.
After making the change and rebooting, I no longer see Piper or Whisper erroring and cycling.
This was the solution to the problem starting Piper (and Whisper) that I opened this thread on, so I am feeling like I need to mark the solution and close this thread now…
However, the next step that I thought was failing due to this problem is actually something else now.
If anyone has a thought here it would be appreciated.
Now that I have both Piper and Whisper running, I nav to Settings > Voice Assistants > “+ Add Assistant” so I can add it and start configuring.
However, still neither Piper or Whisper are available in the drop-down menus to select the engine for TTS and STT. STT only has “None” and TTS has None or Google.
Is there some other step I am missing here in order to make those apps available for voice assistant configuration? I followed the guide I put in my OP and it didn’t mention any other steps.
This post helped me figure it out. Wyoming protocol was not installed, since it wasn’t one of the steps in the guide in my OP and I’m a noob and have no idea how it all works together.
