Year of the Voice - Chapter 2: Let's talk

Wait, does this means you will force https in the app ?, meaning noone with local installation can use it ?, or have to run HA over https ?

For me it works fine on my android, i have not configured SSL either places.

In my opinion, it will work so that the microphone will communicate directly from assist and not through the browser.

2 Likes

ahh ok, make sense ( well if i don’t have to install SSL , that is ) :slight_smile:

Did you find a way to install multiple voices?

Any way to include multiple voices via command:?

Hi :slight_smile:
Any plans for a wyoming-whisper GPU variant/tagged docker image?
Thanks

When I try to manually add the Wyoming Integration I get the message
“Failed to connect”
It might be that I have not input the correct Host and Port.
Should the host be simply the IP address or localhost and what port should I specify?

My docker-compose file includes:-

    whisper:
      container_name: whisper
#    command: --model medium --language nl
      command: --model medium --language en-us
      image: rhasspy/wyoming-whisper
      volumes:
        - ./whisper:/data
      environment:
        - TZ=Pacific/Auckland
      restart: unless-stopped
      ports:
        - 10300:10300

    piper:
      container_name: piper
      image: rhasspy/wyoming-piper
#    command: --voice nl-nathalie-x-low
      command: --voice en-us-ryan-low
      volumes:
        - ./piper:/data
      environment:
        - TZ=Pacific/Auckland
      restart: unless-stopped
      ports:
        - 10200:10200

and I get the following:-

robert@fishhass:~$ docker-compose pull
Pulling homeassistant ... done
Pulling whisper       ... done
Pulling piper         ... done
robert@fishhass:~$ docker-compose up -d
home-assistant is up-to-date
Recreating whisper ... done
Recreating piper   ... done

The ports are as shown in your compose file. The IP will be the IP of the host running docker.

Thanks OzGav.
I was trying using 10.1.1.12 on port 10300
I just tried 10.1.1.12:10200 and now Piper works
but I cannot get 10.1.1.12:10300 to work (for whisper)

Check the logs to see if Whisper is actually starting. I had issues finding a combination of model and language that worked. Edit: you might want to drop the model size back as well otherwise you need to be running on a fast machine

I still have the same problem (on my working machine and on a test machine).

robert@fishhass:~$ docker-compose config
services:
  homeassistant:
    container_name: home-assistant
    environment:
      TZ: Pacific/Auckland
    image: homeassistant/home-assistant:stable
    network_mode: host
    privileged: true
    restart: always
    volumes:
    - /srv/homeassistant:/config:rw
  piper:
    command: --voice en-us-ryan-low
    container_name: piper
    environment:
      TZ: Pacific/Auckland
    image: rhasspy/wyoming-piper
    ports:
    - published: 10200
      target: 10200
    restart: unless-stopped
    volumes:
    - /home/robert/piper:/data:rw
  whisper:
    command: --model medium --language en
    container_name: whisper
    environment:
      TZ: Pacific/Auckland
    image: rhasspy/wyoming-whisper
    ports:
    - published: 10300
      target: 10300
    restart: unless-stopped
    volumes:
    - /home/robert/whisper:/data:rw
version: '3'

Again what is in the docker log? Edit: in fact if you look up you will see that medium English was reported not to work

What do you suggest that I could try?

Here is the latest log after docker-compose restart

robert@fishhass:~$ docker-compose logs
home-assistant   | 2023-05-08 16:43:00.522 ERROR (MainThread) [homeassistant.helpers.entity] Update for camera.tempcam fails
home-assistant   | Traceback (most recent call last):
home-assistant   |   File "/usr/src/homeassistant/homeassistant/helpers/entity.py", line 554, in async_update_ha_state
home-assistant   |     await self.async_device_update()
home-assistant   |   File "/usr/src/homeassistant/homeassistant/helpers/entity.py", line 784, in async_device_update
home-assistant   |     await coro
home-assistant   |   File "/usr/local/lib/python3.10/concurrent/futures/thread.py", line 58, in run
home-assistant   |     result = self.fn(*self.args, **self.kwargs)
home-assistant   |   File "/usr/src/homeassistant/homeassistant/components/zoneminder/camera.py", line 60, in update
home-assistant   |     self._attr_is_recording = self._monitor.is_recording
home-assistant   |   File "/usr/local/lib/python3.10/site-packages/zoneminder/monitor.py", line 169, in is_recording
home-assistant   |     return int(status) == STATE_ALARM
home-assistant   | ValueError: invalid literal for int() with base 10: 'false'

Can we get the voice of GladOS as a text to speach agent?

8 Likes

No I mean you need the logs for the whisper container

For me the medium model would not work, try this:

  whisper:
    command: --model medium-int --language en

My language (Hebrew) does not appear in Piper
I have google translate TTS service but I can’t use it,
In the TTS options in the voice assistant, it is not selectable,
Does anyone know why this is so?
In addition, up to the present time, the SST service has worked well in Assist
But now it doesn’t work properly, what changed?

I have the assist icon which appeared on top right after the update from 2023.4.6 to 2023.5.2 even if I don’t have in my yaml configuration conversation:, default_config: or assist_pipeline:

I was wondering if the assist integration could be disabled or if it cannot be starting 2023.5.x ?

I don’t even have the assist icon on my companion app, I hid the more info button aka ha-button-menu I think, which is why but is it possible to just show the mic and not the button menu instead so I can use it?