Cloning of the sample voice is also not working for me, here is the code:

from TTS.api import TTS
tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2")
tts.tts_to_file( text="My name is Alex, and I'm now on the way to school! I'm going to be talking about some really cool ideas for building colonies on Mars. We're planning to start our journey next year and build an amazing team with our friends. If you're interested in joining us or have any questions, please don't hesitate to reach out. I'm always happy to help! How can I help you today?", speaker_wav="my/sample.wav", language="en", file_path="output/xtts_v2.wav" )

Its result is a mix of different speakers they are speaking the parts of the string, but none of them is the clone of my sample voice.

When I try to use the model yout_tts then the result is same but only with one speaker, again nothing is cloned.

What is wrong in my code, please not, I'm using Mac M1 without GPU.

Which version do you have? There was indeed a bug with inconsistent voice cloning in coqui-tts 0.25.0, but it should be normal in other versions. For better quality, you could fine-tune the model.

Ok, then it's just random variation of the model. For fine-tuning: https://coqui-tts.readthedocs.io/en/latest/models/xtts.html#advanced-training

I've installed the last version two weeks ago. But, as I see, the last version is v0.22.0 (Dec 12, 2023), why are you mentioning 0.25.0?

This repo is abandoned, but we continue to maintain a fork here: https://github.com/idiap/coqui-ai-TTS