Got my M5 Atom Echo, tried to use it for my ChatGPT pipeline, input (whisperer) works great, but the (often long) answer through TTS piper never works beyond the first response sentence or so. Looking at assist debug window the response is sent to piper, it responds (strangely says 0 seconds), but the audio plays only a bit.
Even weirder is that when I try to use TTS Speak service manually afterwards and ask it to say “Hello” using piper to play that on some random speaker in my house, it actually continues the ChatGPT response instead.
Something must be very wrong, any pointers how should I check further please?
Thanks.
Im using
Home Assistant 2023.6.3
Supervisor 2023.07.1
Operating System 10.3
and running whisperer and piper inside docker containers on a different host.