for exmpale creating a a voice in voce editor like this:

Talia speaks at an average pace with an animated delivery in a very confined sounding environment with clear audio quality.

shold use the same speaker every time but that prompt semms to be ignored

Hi @Larrycmd This is (to my knowledge) how Parler works. It literally generates a "new" voice each time. There is little to no-consistency in the voices. Other people comment on such an issue e.g. huggingface/parler-tts#112

I have even asked the question about consistency of the supposed "built in voices" huggingface/parler-tts#96

I also note this in the Engine information:

To be clear, AllTalk is handing over the text you input to Parler's own scripts and its their own code that is generating the TTS. AllTalk is not manipulating/changing or altering anything and it is using Parler's suggested code implementation. I would suggest you add any comments/suggestions to the people at Parler https://github.com/huggingface/parler-tts and they can look to implement changes or better explain how the engine works (or doesnt work).

If so, How to do?

The whole Process is XTTS. Just all the steps needed.

i dont have it

what should i do

Use the start_finetune script to start it. Not start_alltalk

I want finetune vits model, 😓

run with many error