F-0310: Standalone local inference setup for XTTS-v2 + Qwen 2.5 on Windows (No Cloud/API)

Beginners
1

Hey everyone :waving_hand:

For those working with Coqui XTTS-v2 and local safety guardrails on Windows, I built a zero-dependency, fully local deployment wrapper named F-0310.

:hammer_and_wrench: Key Pipeline Details:

  • Safety & Prompt Filtering: Integrated Qwen 2.5 to run local guardrails before passing prompts to speech synthesis.
  • Audio Core: Coqui XTTS-v2 for high-fidelity zero-shot voice cloning.
  • Windows Automation: Ships with clean, inspectable .bat scripts (install.bat / run.bat) that set up isolated environments automatically without compiled/obfuscated binaries.
  • Hardware Detection: Dynamically switches execution backends between NVIDIA (CUDA 12.1), AMD DirectML, Intel Iris/Arc, or CPU fallback.

(Full architecture breakdown and benchmark notes are provided in the comments below)

I’d love to hear feedback from anyone running this across different GPU environments!

IMG_0745
IMG_07451375×768 227 KB