Local VA Pipeline - No Addons

[steep otter]

I'm looking for walkthrough recommendations to run a local (networked) voice assistant pipeline. Most (all?) of the HomeAssistant docs/results point towards "just click Add-ons," but I'm running HA on docker so I don't have that option.

I've managed to make progress through quite a few attempts, but troubleshooting pytorch and cuda finally made me stop and humbly post here for help. I'd love to see a walkthrough or list of how to get whisper (which one 😦 faster-whisper? wyoming-faster-whisper? purfview?) and piper running on this GPU-accellerated host (not picky about OS).

Very glad to provide more info and appreciate any helpful thoughts. 🤘

[lusty lintel]

you want to get whisper and piper containers that have wyoming support. that will allow you to connect them to HA using the wyoming integration

[steep otter]

thanks - yes. that's definitely the route that i've been heading down. specifically, i am having trouble getting whisper to use the GPU to run faster in that scenario

[lusty lintel]

I have had good luck with the linuxserver container - https://docs.linuxserver.io/images/docker-faster-whisper/#docker-compose-recommended-click-here-for-more-info

heres a cat of my docker compose file for it, if it

version: '3'

services:
  faster-whisper:
    image: lscr.io/linuxserver/faster-whisper:gpu
    container_name: faster-whisper
    environment:
      - PUID=1000
      - PGID=1000
      - TZ=Etc/UTC
      #- WHISPER_MODEL=tiny-int8
      #- WHISPER_MODEL=base.en
      #- WHISPER_MODEL=medium-int8
      - WHISPER_MODEL=cstr/whisper-large-v3-turbo-int8_float32
      - WHISPER_BEAM=5 #optional
      - WHISPER_LANG=en #optional
    volumes:
      - ./config/:/config
    ports:
      - 10300:10300
    restart: unless-stopped
    devices:
      - /dev/dri:/dev/dri
    deploy:
      resources:
        reservations:
          devices:
          - driver: nvidia
            count: all
            capabilities: [gpu]

[steep otter]

beautiful - i'll run this today to give it a shot. it looks like it manages all of those pieces that i was having trouble with. thank you SO much, @lusty lintel

[lusty lintel]

no worries. hopefully you get up and running in no time 🙂

[steep otter]

right? and then on to the next troubleshooting adventure! 😄 i appreciate your help

[steep otter]

ayyyy