Buzz Mini 1.0.1

Local offline push-to-talk dictation for Windows (faster-whisper + NVIDIA CUDA). Whisper model weights are downloaded on first use from the Models tab.

Install (Windows)

1. BuzzMini-Setup-1.0.1.exe — small installer; downloads the app bundle from this release during setup (~4+ GB after extract).
2. Or download BuzzMini-1.0.1-win64.7z and extract manually.

GPU: NVIDIA + driver required for GPU transcription. AMD / Intel GPUs run on CPU only — use a smaller model (Small / Base) for speed.

Docs: README (English) · README.ru (Русский)

---

What's new since 1.0.0 (2729794)

Push-to-talk (PTT)

• Fixed chord handling on Windows: reliable left/right Ctrl + Space / Win via virtual-key codes.
• PTT model: hold Ctrl, press the second key to start recording (fewer false triggers).
• Hotkey listener restarts after sleep / lock screen so PTT keeps working.

GPU / CPU

• Detects AMD (and similar) GPUs via WMI; clear message that CUDA needs NVIDIA.
• Separate log warnings for CPU-only PyTorch vs CUDA build without a usable GPU.
• Settings shows transcription device status; tray notification on first run in CPU mode.

Installer

• Payload download via PowerShell (GitHub HTTPS / redirects, large .7z files).
• Clearer setup text about download size and disk space.

Docs

• README.md (English) + README.ru.md (Russian) with language switch at the top.

Assets

File	Description
BuzzMini-Setup-1.0.1.exe	Web installer (recommended)
BuzzMini-1.0.1-win64.7z	Portable PyInstaller bundle