Skip to content

Version 1.1.0 - Major Rewrite, to allow modular appoach and LuxTTS support

Choose a tag to compare

View all tags
@FranckyB FranckyB released this 08 Feb 02:45
· 83 commits to main since this release
v1.1.0
c80e845
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Voice Clone Studio v1.0.0 —

I'm thrilled to announce Voice Clone Studio v1.0.0 — a complete ground-up rewrite that transforms the project from a 6,000+ line monolith into a clean, modular architecture. This has been a quite the effort and I'm excited to share it!

What's New

Fully Modular Architecture

The entire application has been rewritten. Every tab is now a self-contained tool module that lives in its own file, manages its own UI, and handles its own events. The main file? It's now just ~230 lines of orchestration.

This means:

  • Each tool can be tested independently in standalone mode
  • Adding new tools is as simple as creating a single file and registering it — no touching the main app

🔧 Customizable Tool Visibility

Not using Voice Design? Don't need the Train Model tab? Now you can turn tools on and off right from Settings! Each tool can be toggled independently, keeping your workspace clean and focused on what you actually use. Your preferences are saved and persist between sessions.

📂 Brand New FileLister Component

I built a custom Gradio component that makes file management a joy:

  • 🖱️ Double-click to play — Click any audio file and it plays instantly, no extra steps
  • ☑️ Multi-select — Select multiple files at once for batch deletion
  • Clean, responsive design that fits more naturally into the UI

🎛️ Improved Settings

Settings got an upgrade:

  • Visible Tools — Toggle any tab on or off
  • Help Guide — The help documentation now lives inside Settings as a clean sub-tab, freeing up space in the main tab bar
  • Settings icon pushed to the far right of the tab bar for quick access ⚙️

🏗️ Under the Hood

For the technically curious, here's what changed behind the scenes:

  • AI Model Managers — Centralized TTS and ASR model management with automatic VRAM optimization. Models unload intelligently when switching engines.
  • Shared State Architecture — Tools receive everything they need through a unified shared state dictionary — config, utilities, model managers, UI helpers. No globals!
  • Centralized Constants — All model sizes, languages, speakers, and generation defaults defined once in a single constants.py. No more hunting for duplicated values.
  • Emotion Manager — The 40+ emotion preset system is now its own module with clean save/delete/reset workflows
  • Audio Utilities — Normalize, denoise (DeepFilterNet), mono conversion, format checking — all extracted into reusable utilities

🎵 Prep Audio (formerly "Prep Samples")

The sample preparation tool now serves dual purpose:

  • Samples mode — Prepare voice samples for cloning (same great workflow)
  • Datasets mode — Manage finetuning datasets with batch transcription

One unified tool, two powerful workflows.

📊 By the Numbers

Before (v0.7.6) After (v1.0.0)
1 file, 6,000+ lines 8 tool modules + core libraries
Globals everywhere Shared state architecture
Monolithic UI Independent, testable tools
Fixed tab bar Configurable tool visibility
Basic hacked file lists Custom FileLister with multi-select & playback

🚀 Getting Started

Nothing changes for users — just pull the latest and launch as usual! Your existing config.json, samples, datasets, and trained models all carry over seamlessly.

# Windows
launch.bat

# Linux/Mac
./launch.sh

Happy cloning! 🎤✨

It's up in Dev now and should be pushed to Main soon.

Assets 2
Loading