Free local Whisper voice-to-text app for Windows and Linux with desktop launchers, local model cache, a working Windows installer, and Windows NVIDIA GPU support

[Bamamana]

I have been working on a free local Whisper voice-to-text app that runs from one repo on both Windows and Linux.

Repo:
https://github.com/Bamamana/whisper-voice-to-text

What it currently does:

• microphone recording
• audio and video file transcription
• desktop shortcut or icon setup on both Windows and Linux
• local model caching in model-cache/
• optional model pre-downloads so switching models is faster later
• Windows NVIDIA GPU support with CUDA runtime libraries installed into the app environment
• automatic CPU fallback if CUDA is unavailable

Windows installer:

• the packaged Windows installer is WhisperVoiceToTextSetup.exe
• if you build it from the repo, run build_windows_installer.bat
• the generated installer is written to dist\windows-installer\WhisperVoiceToTextSetup.exe
• the installer copies the app into %LOCALAPPDATA%\Programs\Whisper Voice To Text, creates Start Menu shortcuts, can create a desktop shortcut, and runs the Windows setup flow for you
• on a clean Windows machine it can bootstrap Python and download an app-local FFmpeg build automatically when they are missing

If someone wants the repo-based Windows path instead of the packaged installer, they can still just run setup_windows.bat from the repo folder.

What I focused on:

• making the repo usable by non-developers
• clear setup docs for Windows and Linux
• keeping everything local and offline after the initial setup and model downloads
• making models download once and then be reused locally

Main docs:

• README.md
• QUICKSTART.md
• FIRST_RUN.md
• WINDOWS_NOTES.md