Feature Request: Integrate GPT-SoVITS as TTS Backend Option

[GAODBK]

Hi, thank you for your great work on this project!

I'd like to suggest a new feature: supporting the [GPT-SoVITS](https://github.com/RVC-Boss/GPT-SoVITS) text-to-speech system as an optional backend for audiobook generation.

Why GPT-SoVITS?

• It produces high-quality, expressive speech that is more human-like than traditional TTS engines.
• It supports multi-speaker voice cloning, which could be useful for customizing the audiobook narrator's voice.
• It has an active community and provides pretrained models via Hugging Face:
  https://huggingface.co/lj1995/GPT-SoVITS/tree/main

Proposal

• Add a new --tts_engine gpt-sovits CLI flag to select GPT-SoVITS.
• Allow specifying model paths and speaker IDs as arguments.
• Optionally provide integration with [VoiceConversionWebUI](https://github.com/RVC-Boss/GPT-SoVITS) or expose a simplified interface to run inference in the background.

This would open up many possibilities for personalized and high-fidelity audiobook generation.

Let me know what you think — happy to help test or experiment with integration if needed.

Thanks again for this awesome tool!

[ROBERT-MCDOWELL]

it's on the roadmap list. I will contact you when it will be time to do it. thanks!