#Problems using a custom voice in Piper Add-on

1 messages · Page 1 of 1 (latest)

analog orchid
#

I have a custom voice that I’m using for my assistant. I followed instructions and added both required files into the share/piper folder via Samba share.

I used Data’s voice from Star Trek: https://github.com/dividebysandwich/piper-voice-models

I renamed both files from “en_US-data_7024-medium” to “lt_cdr_data” with all relevant extensions in place.

The custom voice is selectable in the Pipeline configuration and works fine when it responds through both my VACA Echo Show 5 and my Voice PE. All voice responses use the custom voice.

But I can’t set it as the default voice in the piper add-on. It’s just not in the list. As such, a few automations that run calls through Piper directly end up using the default lessac voice instead. I’ve already tried reloading, restarting, a full shutdown, etc. now I’m at a loss.

GitHub

Pre-trained and ready to use voice models for Piper TTS - dividebysandwich/piper-voice-models

wise brook
#

renaming it is not usually a great plan and can cause issues. although it might not be the issue here to be fair, its probably best to start with a known default.

analog orchid
#

That was a thought I had, but it works in the Voice Assistant config and works just fine.

It’s just not showing up in the add-on config to be set as the system default.

wise brook
#

also in your automation if you use
assist_satellite.announce instead of tts.speak then it should use the default voice of the pipeline assigned to the device.