Drop-in replacement proxy for Voiceserver: Voiceproxy

[ryhanen]

First of all, I must say that I'm really grateful for the work @danielmiessler and other contributors are doing with the development of the PAI scaffolding. I have sort of been avoiding the whole AI topic until couple of weeks ago I decided to dive into it. And boy, did I dive deep. I'm running PAI in a WSL2 environment and it has taken some tinkering with my PAI to get things going smoothly. But at the same time PAI has taken me back to "hands-on" software development after few quiet years. Although. this time "hands-on" means something different.

While I have been able to get the Voiceserver working in my WSL2 environment (with help from PAI), I have felt that doing upstream changes like that will be never ending headache considering the pace that things are evolving. At the same time I have a pretty extensive Home-Assistant setup with existing TTS functionality at my home which enables various advanced scenarios when it comes to voice announcements (based on presence, whether I'm in a meeting, etc.). So I started to think if I could improve the situation.

Another driver for this project was the fact that I'm not native english speaker. And if I want to speak to my PAI in my native Finnish language, the TTS announcements will sound off if I can't express the correct language. Voiceproxy makes it possible for PAI to indicate the language that the TTS service should use.

While I think that in the long run it would be better to do a more clean separation of the Voiceserver and PAI in the upstream codebase, I started with doing what would cause the least amount of issues with the upcoming PAI releases (apart from making PAI multi-language aware).

This project is an drop in replacement for PAI's default Voiceserver. My use case is to re-route the voice announcements to my Home-Assistant via automation triggered with a webhook. All code and documentation is generated by PAI, my role was more on the supervising side. Amazing times.

Voiceproxy repo

[KimVatnedal]

Great work. My DA added local playback as well, so now my request goes to Elevenlabs and plays on my computer.

[danielmiessler]

Very cool, thank you!