Can anyone recommend any free half decent TTS that is compatible with home assistant that does not run out after a few days and is quick.
Thanks
#Free TTS
pure blaze
slow igloo
half decent is a bit subjective. I for instance use Piper with one of the medium/high models and it works plenty fine for me, and is nearly instant in its responses. Some people also use Kokoro TTS though that requires some advanced setup and a GPU I think.
long bane
Can anyone recommend any free half decent TTS that is compatible with home assis...
How many characters per month do you require?
https://www.home-assistant.io/integrations/amazon_polly/
https://github.com/hugobloem/wyoming-microsoft-tts
The Piper container with acceleration on the gpu seems to be the only suitable local option for talking to LLMs.
When streaming support becomes available, the situation will change
karmic willow
Does Piper benefit from GPU acceleration? If so, could it use the RPi GPU, or is it limited by memory bandwidth rather than by comptue?
slow igloo
Supposedly there's dockers that can make use of GPU, but even on a modest system with a decent CPU (like an intel i5 or something in one of those NUCs) the performance is quite good. I am running a GPU accelerated version, but I don't really notice much if any speed improvement over running it on my CPU.