Docker: HA + Piper + Whisper - Try TTS: "Error Playing Audio"

[pseudo forge]

Created docker containers for HA, Piper, Whisper. Added Piper and Whisper successfully to HA. When I attempt to "test" the TTS, it just spins for a few minutes, and I get the message, "Error Playing Audio". Logs for HA and Piper show no errors. Any thoughts?

Docker: HA + Piper + Whisper - Try TTS: "Error Playing Audio"

[livid moth]

can you share your compose files?

[pseudo forge]

---
networks:
  zinternal0:
    external: true
    name: zinternal0

volumes:
  ha0_config:
    name: ha0_config
    driver_opts:
      type: nfs
      o: "addr=*.*.*.*,nolock,soft,rw"
      device: ":/path/to/data/ha0/config"
  piper0_config:
    name: piper0_config
    driver_opts:
      type: nfs
      o: "addr=*.*.*.*,nolock,soft,rw"
      device: ":/path/to/data/piper0/config"
  whisper0_config:
    name: whisper0_config
    driver_opts:
      type: nfs
      o: "addr=*.*.*.*,nolock,soft,rw"
      device: ":/path/to/data/whisper0/config"

services:
  homeassistant:
    image: lscr.io/linuxserver/homeassistant:latest
    container_name: ha0
    restart: unless-stopped
    networks:
      - zinternal0
    volumes:
      - ha0_config:/config
      - /etc/localtime:/etc/localtime:ro
    environment:
      - PUID=1000
      - PGID=1000
      - TZ=America/New_York

  piper:
    image: lscr.io/linuxserver/piper:latest
    container_name: piper0
    environment:
      - PUID=1000
      - PGID=1000
      - TZ=America/New_York
      - PIPER_VOICE=en_US-lessac-medium
      - PIPER_LENGTH=1.0 #optional
      - PIPER_NOISE=0.667 #optional
      - PIPER_NOISEW=0.333 #optional
      - PIPER_SPEAKER=0 #optional
      - PIPER_PROCS=1 #optional
    volumes:
      - piper0_config:/config
    networks:
      - zinternal0
    ports:
      - 10200:10200
    restart: unless-stopped

  faster-whisper:
    image: lscr.io/linuxserver/faster-whisper:latest
    container_name: whisper0
    environment:
      - PUID=1000
      - PGID=1000
      - TZ=America/New_York
      - WHISPER_MODEL=tiny-int8
      - WHISPER_BEAM=1 #optional
      - WHISPER_LANG=en #optional
    volumes:
      - whisper0_config:/config
    networks:
      - zinternal0
    ports:
      - 10300:10300
    restart: unless-stopped

[livid moth]

ok, so first thought is:
why are you using the linuxserver image instead of the official image?

but i guess more to the point, does your network setup allow for the containers to connect to the internet (to get the models)?

[pseudo forge]

linuxserver img because it comes with a config, easy copy/paste. that's where I get most of my docker-configs from.

"zinternal0" can reach out to the web.

admittedly it's not the best naming for the network, but for what I needed to do at the time it made sense.

[livid moth]

I definitely suggest moving to the official image there are compose files available for it too if that's what you need.

so does the wyoming integration seem to connect fine but it just hangs/times out when you try and use it?

[pseudo forge]

Correct, that is the behavior I am experiencing.

[pseudo forge]

I'm on the github and only see default docker script, not compose.

I'm not the best with knowing how to convert that to docker-compose, hence why I typically use linuxserver.io

[livid moth]

https://www.home-assistant.io/installation/linux#docker-compose

[pseudo forge]

nice, you rock.

thx

[livid moth]

also, since 2025.8 there has been a bug with HA and TTS agents that don't support "streaming" that could be part of the issue

[pseudo forge]

Oh I think I misunderstood- you meant HA for docker compose?

[pseudo forge]

ofc

haha

[livid moth]

a fix has been merged and its added to the milestone for release in 2025.8.2. that may help but it may not

[livid moth]

home assistant for starters but all of the images can be moved over to the stock ones.

[livid moth]

here is a secret weapon for you 🙂 https://www.composerize.com/

if you ever want to be lazy and convert docker run commands to compose that will do it

so for example for piper it would be

services:
    wyoming-piper:
        stdin_open: true
        tty: true
        ports:
            - 10200:10200
        volumes:
            - /path/to/local/data:/data
        image: rhasspy/wyoming-piper
        command: --voice en_US-lessac-medium

[pseudo forge]

Nice thanks. I'll revisit this maybe in a few days or so to see if anything has changed with the fix.

Meanwhile I'll also be converting to the official containers.

[pseudo forge]

I converted to official HA docker-compose, 2025.8.1, piper still doesn't work. I guess I just have to wait.

Another thing to note, it seems in all the video guides to set up piper/whisper, there's some kind of config area within Settings > Devices > Integrations > Piper/Whisper. I do not have this. I have Settings > Devices > Integrations > Wyoming Protocol > Piper/Whisper. In there, I don't see any configuration options to speak of, other than "expose" which I enabled. (which I now know is pointless)

edit: nvm, these settings are already being configured in my docker-compose file.