Starstruk Logo MarkStarstruk

v4.0.0

vercel/streamdownv4.0.0Jul 13, 2026by abus-aikorea

AI Summary

Voice-Pro v4.0.0 fully opens the project to the community and removes all costs by modernizing the installation stack and AI engines. The update introduces a fast, portable installation method using `uv` and upgrades the runtime environment.

Key Highlights

  • Installer migrated to `uv` for fast, reproducible, and portable installation
  • No administrator rights, CUDA Toolkit, or Visual Studio Build Tools required
  • Upgraded to Python 3.12, Torch 2.8.0 (RTX 50-series supported), and Gradio 6.20
  • New optional TTS model: Fun-CosyVoice3-0.5B
  • Enhanced UI with full-width layouts and visible error toasts

Breaking Changes

  • Removal of `whisperX` due to dependency conflicts with Gradio 6
  • Removal of CUDA Toolkit and Visual Studio Build Tools requirements
  • Migration path requires deleting `installer_files/` folder to use new `uv` lockfile

New Features

  • Open source and completely free
  • Azure Translator and TTS support via `.env` configuration
  • Self-healing model downloads for interrupted transfers
  • Silent uninstallation capability
  • Subtitle tracks shown directly in video players

Full Release Notes

# Voice-Pro v4.0.0

Voice-Pro is now fully open source and completely free. This release rebuilds the entire installation stack and modernizes every AI engine.

## ⚡ Highlights

- **Installer migrated from Miniconda/pip to [uv](https://docs.astral.sh/uv/)** — installation now takes **minutes, not hours**, and is fully reproducible from the committed `uv.lock`. Everything stays inside `installer_files/` (uv, Python, packages) — a truly portable install.
- **No administrator rights needed** — `configure.bat` is now optional. `start.bat` auto-downloads a portable **ffmpeg** if it is not installed. CUDA Toolkit and Visual Studio Build Tools are **no longer required** (PyTorch bundles the CUDA runtime; all dependencies ship prebuilt wheels).
- **Upgraded runtime**: Python 3.12, Torch 2.8.0+cu128 (**RTX 50-series supported**), Gradio 6.20.

## 🎙️ Latest AI engines

- **ASR**: faster-whisper 1.2.1 (large-v3-turbo, distil-large-v3.5), openai-whisper 20250625, whisper-timestamped 1.15.9. whisperX was removed (its dependency pins blocked the Gradio 6 upgrade; existing configs fall back to faster-whisper).
- **TTS**: F5-TTS 1.1.21, kokoro 0.9.4, edge-tts 7.x, and re-vendored CosyVoice (upstream main).
- 🇰🇷 **New optional TTS model: Fun-CosyVoice3-0.5B** — 9 languages including Korean, selectable in the CosyVoice tab (downloads from the official HF repo on first use).

## 🛡️ Restricted / corporate PC friendly

- Whisper model downloads self-heal after interrupted/corrupted transfers.
- Translation automatically retries with backoff when the network rate-limits the free Google endpoint (failed lines are reported, originals kept).
- Optional **Azure Translator / Azure TTS** with your own keys via `.env` — see the "Azure services" section in the README.

## 🖥️ UI improvements

- Migrated to **Gradio 6**: full-width layout for all tabs, subtitle tracks shown directly in the video players.
- **Errors are now visible**: every failure shows a red error toast that stays on screen until you close it, with actionable messages for common causes (missing ffmpeg, no media registered, etc.).
- `uninstall.bat` no longer requires administrator rights and no longer force-reboots; `uninstall.bat silent` runs unattended.

## 📀 Install / Upgrade

- **Fresh install**: clone or download the source, then run `start.bat`. First run installs uv + Python 3.12 + all dependencies from the lockfile (minutes), then downloads AI models (~10GB — this is the slow part).
- **Upgrading from v3.x**: delete the `installer_files/` folder and run `start.bat` again. Your `model/` and `workspace/` folders are kept.

**Full Changelog**: https://github.com/abus-aikorea/voice-pro/compare/v3.2.0...v4.0.0