[BUG] Cannot add custom voices

[wookash]

Is there an existing issue for this?

• I have searched the existing issues

Current Behavior

I'm adding new voice onnx and relevant onnx.json files to the config folder, I'm updating voices.json, the voice can be chosen in home assistant but there is error (see below)

Expected Behavior

I'd like to be able to add custom voice

Steps To Reproduce

copy voice files to config directory
update voices.json file
reload wyoming integration in home assistant
choose new voice in assistant
click play to try the new voice
error occurs

Environment

- OS: debian 12
- How docker service was installed: portainer

CPU architecture

x86-64

Docker creation

services:
  piper:
    image: lscr.io/linuxserver/piper:latest
    container_name: piper
    environment:
      - PUID=xx
      - PGID=xx
      - TZ=xxx/xxx
      - PIPER_VOICE=en_US-lessac-high
      - PIPER_LENGTH=1.0 #optional
      - PIPER_NOISE=0.667 #optional
      - PIPER_NOISEW=0.333 #optional
      - PIPER_SPEAKER=0 #optional
      - PIPER_PROCS=1 #optional
    volumes:
      - /xxxx:/config
    ports:
      - 10200:10200
    restart: unless-stopped

Container logs

ERROR:asyncio:Task exception was never retrieved

future: <Task finished name='Task-9' coro=<AsyncEventHandler.run() done, defined at /lsiopy/lib/python3.12/site-packages/wyoming/server.py:28> exception=VoiceNotFoundError('jarvis')>

Traceback (most recent call last):

  File "/lsiopy/lib/python3.12/site-packages/wyoming/server.py", line 35, in run

    if not (await self.handle_event(event)):

            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "/lsiopy/lib/python3.12/site-packages/wyoming_piper/handler.py", line 73, in handle_event

    piper_proc = await self.process_manager.get_process(voice_name=voice_name)

                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "/lsiopy/lib/python3.12/site-packages/wyoming_piper/process.py", line 114, in get_process

    ensure_voice_exists(

  File "/lsiopy/lib/python3.12/site-packages/wyoming_piper/download.py", line 77, in ensure_voice_exists

    find_voice(name, data_dirs)

  File "/lsiopy/lib/python3.12/site-packages/wyoming_piper/download.py", line 183, in find_voice

    raise VoiceNotFoundError(name)

wyoming_piper.download.VoiceNotFoundError: jarvis

[github-actions]

Thanks for opening your first issue here! Be sure to follow the relevant issue templates, or risk having this issue marked as invalid.

[LinuxServer-CI]

This issue has been automatically marked as stale because it has not had recent activity. This might be due to missing feedback from OP. It will be closed if no further activity occurs. Thank you for your contributions.

[rodrigosiviero]

same here, @wookash have you been able to solve it?

[wookash]

same here, @wookash have you been able to solve it?

Nope, I gave up for now - running other projects..

[LinuxServer-CI]

This issue has been automatically marked as stale because it has not had recent activity. This might be due to missing feedback from OP. It will be closed if no further activity occurs. Thank you for your contributions.

[NeoJoris]

Any updates? @wookash @rodrigosiviero

I would like to add Jarvis voices from https://huggingface.co/jgkawell/ (atleast be able to try custom voice (links))

[rodrigosiviero]

Any updates? @wookash @rodrigosiviero

I would like to add Jarvis voices from https://huggingface.co/jgkawell/ (atleast be able to try custom voice (links))

Yes, I've been able to use my custom onnx voice but not with this image, with HA addon itself like this:

  piper:  
      image: homeassistant/amd64-addon-piper:latest  
      container_name: piper
      entrypoint: python3  
      command: >  
        -m wyoming_piper  
        --piper '/usr/share/piper/piper'  
        --uri 'tcp://0.0.0.0:10200'  
        --length-scale "2"  
        --noise-scale "0.667"  
        --speaker "0"  
        --voice "en_GB-jarvis-high" 
        --max-piper-procs "1"  
        --data-dir /data  
        --data-dir /share/piper  
        --download-dir /data  
      ports:  
        - "10200:10200"  
      volumes:  
        - /piper/data:/data  
      restart: unless-stopped

1. Mount /data on the container to the host
2. Open the file named voices.json
3. Add the block of code for your specific onnx and onnx.json
4. generate the MD5SUM from the files and size in bytes
5. Upload the files at /data directory(root itself, no need to use folders just copy everything there)
6. restart piper/ha just to be sure and select on HA

here's the block for jarvis, I already did it for you

    "en_GB-jarvis-high": {
        "key": "en_GB-jarvis-high",
        "name": "Jarvis",
        "language": {
            "code": "en_GB",
            "family": "en",
            "region": "GB",
            "name_native": "English",
            "name_english": "English",
            "country_english": "Great Britain"
        },
        "quality": "high",
        "num_speakers": 1,
        "speaker_id_map": {},
        "files": {
            "en/en_GB/jarvis/high/en_GB-jarvis-high.onnx": {
                "size_bytes": 114199011,
                "md5_digest": "2921a440ebf4064ec761a5a02f66c65a"
            },
            "en/en_GB/jarvis/high/en_GB-jarvis-high.onnx.json": {
                "size_bytes": 7262,
                "md5_digest": "fecff68ce6f9b52ffaae52f06d824b05"
            },
            "en/en_GB/jarvis/high/MODEL_CARD": {
                "size_bytes": 309,
                "md5_digest": "b116c3cbdebac99ade9af03807cb9301"
            }
        },
        "aliases": [
            "en-gb-jarvis-low"
        ]
    },

note, I dont think the MODEL_CARD matters, I just left it there the original one that container downloads, or you can download the jarvis one and md5sum it as well.

Damn I wish I had the Jarvis voice in PT_BR

[Ocramius]

FWIW, what stands mostly in the way seems to be --update-voices:

docker-piper/root/etc/s6-overlay/s6-rc.d/svc-piper/run

Line 16 in 39dd7df

--download-dir /config --update-voices

[LinuxServer-CI]

This issue has been automatically marked as stale because it has not had recent activity. This might be due to missing feedback from OP. It will be closed if no further activity occurs. Thank you for your contributions.