v4.0.0 - Leaner, Louder, Global

What's New 🎉

🌍 Multi-Language Support
You can now transcribe audio in over 90 languages! Previously locked to English, the app now lets you pick from Afrikaans to Chinese and everything in between — with a handy language selector right in the settings.

🎛️ Smarter Settings Panel
The model and precision options are now split into separate dropdowns, making it much easier to pick exactly what you want. The app also automatically disables options that don't apply — for example, English-only models will gray out the language and translation controls so you don't have to guess.

🧹 Cleaner Architecture
Under the hood, the app has been significantly simplified. Redundant code was removed, the model configuration was consolidated from 200+ lines of copy-paste into a compact table, and the visualization system (bars, sparklines, speedometers, arc graphs) now shares a single set of definitions instead of repeating itself four times.

⚡ Other Improvements

• The app now talks directly to the transcription engine instead of going through an unnecessary middle layer, making the code easier to maintain.
• Smarter CPU thread detection for better out-of-the-box performance.
• The installer script was tidied up and simplified.
• Whisper large-v3 turbo model added to the lineup. 🚀

v3.2.1 - improved metrics bar

Metrics Bar Reliability & Performance

• Replaced the custom subscriber callback system in MetricsStore with a Qt signal, making cross-thread safety explicit rather than relying on implicit queued connection behavior
• Fixed several resource leaks: MetricsCollector threads are now properly cleaned up on visualization switches, pynvml.nvmlShutdown() is now called on cleanup, and GPU availability is probed once rather than re-initialized per visualization
• Cached all static pixmaps (speedometer arcs, sparkline gradients, arc backgrounds) as instance attributes created at init, fixing silent QPixmapCache.find failures in PySide6 that caused per-frame recreation
• Reduced default polling interval from 200ms to 400ms and misc cleanup including a threading.Event-based stop flag and corrected field naming

v3.2.0 - fast install

Changes

• Windows installer — download and run Whisper_S2T_Transcriber_Setup.exe for one-click setup
• Python 3.13 support
• Simplified installation — single-pass uv pip install replaces multi-phase install; dependency list trimmed to essentials via whisper-s2t-reborn package
• Bug fix — resolved double mutex unlock in model cleanup
• Housekeeping — cleaned up repo (removed tracked __pycache__ files), refreshed README

v3.1.0 - python 3.13 support

• support python 3.13
• upgrade to torch 2.9.0 and CUDA 12.8
• update other dependencies
• minor bug fix

v3.0.0 - major refactor

Major refactoring for future expansion. Also, more reliable functionality.

v2.0.0 - enhanced

1. updated dependencies
2. added .webm file support
3. use a single setup.py installation script instead of separate scripts for GPU versus CPU
4. added Python 3.11 and Python 3.12 support.
5. enhanced metrics bar

You can now right-click on the metrics bar and stop monitoring altogether, if you're concerned about the small overhead...or you can choose between different cool visualizations.

v1.3.3 - WhisperS2T-transcriber

Patch release to make setting cuda-related paths more robust.

v1.3.2 - user-friendliness

Changes

• The comboboxes are now only populated with the device and whisper model precisions that your system supports, and are updated dynamically if you change the compute device.

v1.3.1 - better-faster-stronger

Improvements

• Add checks for ensuring the correct Whisper model is selected depending on whether CPU or GPU is selected.
• Adjust how many CPU cores are utilized for the best performance.

v1.3.0 - better, faster, stronger

Improvements

• Added distil whisper models as well as Large-v3 model variants.
• Significantly improved the installation speed by using the uv library.
• Users now choose the installation method of GPU support or Cpu-only.
• You no longer have to separately install CUDA on your computer. The GPU installation will pip install into the virtual environment.

Removals

• Removed Python 10 support.