I’m still running my custom homeassistant - rhasspy 2.5 setup.
Homeassistant’s wyoming integration slowly gets where my own setup is already but it’s not yet there. So I cannot switch just yet.
However, I wanted to use the better quality of piper TTS for it.
I built a small docker container that runs piper and also provides the basics for MaryTTS web interface. You can use it in rhasspy as a drop in replacement for Mimic3 or MaryTTS. It also automatically downloads the voices if not present and allows to add custom voices as well.
I’m not sure if anybody still runs on rhasspy 2.5 but if you want to try here it is:
Wow Pete, this is exactly what I’ve been wanting to do, but didn’t have the skill set to create. I’m very happy with 2.5, I’m not looking for an Alexa replacement, I just want to control devices in HA and to have announcements. Using Kaldi & Fsitcuffs along with slots works well for part one, but having a really good “voice” was what I’m missing. I set up Mimic3 with the MaryTTS integration, which is better than Larynx, but I get weird clipping on some tts. So, I’ll set this up, and let you know how it goes.
Thank you very, very much!!
Jeff
Hey Pete,
Got this set up, and its working great!!! Thank you so much for doing this, much better voice than Larynx & Mimic (and thank you @synesthesiam for creating Piper!) and I’m able to stay with my 2.5 implementation. On my Rpi4, performance very good, with the longer responses or announcements only taking a second or two to start. Very happy camper!!!
Jeff
