The first time I use the voice assistant with a new phrase (or maybe it's the new response), the audio does not play. I have to ask the question a second time. The whole pipeline appears to work successfully in debug. The only anomaly is that the "Text-to-speech" phase is always "0s" when it definitely takes longer than that to generate. I feel like its just not waiting for it to finish so by the second time I ask, it has finished and it's now cached.
#First text to speech doesnt play
hasty current
viscid isle
Yeah looks like that. What do you use for TTS?
hasty current
Originally I setup a piper-local which is running on the rasp pi 4 b
I then setup a remote piper in a docker container on my desktop but was getting the same thing
of course... I just tried again and I think it's all of a sudden responding the first time with that setup?!?!?.....
though debug still shows 0s for the TTS
viscid isle
There's also HA, proxying the response, involved. So it's not just Piper itself that can glitch, but the HA proxy. I'd say "try moving HA to beefier platform", but i understand how it sounds. 🙂
hasty current
Yea I've thought about that. I really wish I could run the whole pipeline on a single machine. I'm pretty sure ESPHome is even crashing my HAOS right now. I want something small without fans (or at least completely silent ones)