EasyWhisperUI

A fast, local desktop app for transcribing audio/video with Whisper (whisper.cpp) — with GPU acceleration where available.

---

Electron migration (React + Electron + IPC, TypeScript)

EasyWhisperUI has been migrated to an Electron architecture (React + Electron + IPC) built with TypeScript to provide a consistent cross-platform UI (Windows + macOS + Linux) and a faster, more reliable development workflow going forward.

What changed / why it matters

• Unified UI across platforms with consistent layout + behavior
• Faster iteration (React + Vite, easier UI/feature work)
• TypeScript codebase for safer refactors and more reliable IPC/renderer/main boundaries
• Safer process boundaries
  • Renderer UI has no direct Node access
  • Privileged operations live in the Electron main process
  • A narrow preload bridge (window.easyWhisper) handles IPC safely
• Cleaner installs + predictable storage
  • Per-user workspace (app data) keeps binaries/models/downloads organized
• Better foundation for long-term UX improvements

This rewrite required reworking core UI flows, IPC, and install/dependency handling. The Electron build has been tested multiple times on a fresh Windows system to validate clean installs and end-to-end transcription.

---

Features

• Live transcription (beta)
• Batch transcription queue (multiple files processed sequentially)
• Translation support for 100+ languages
• Output formats:
  • .txt
  • .srt (timestamps)
• Drag & drop + Open With integration
• Automatically converts media to the required format via FFmpeg
• Model selection (e.g. tiny, medium-en, large-v3)
• Language selection (e.g. en)
• Optional “additional arguments” textbox
• Automatically downloads models when missing
• Console output view during processing
• Custom model support: select a local whisper.cpp-compatible model file directly via a file picker

---

Acceleration

• Windows: Vulkan acceleration (supported GPUs)
• macOS (Apple Silicon): Metal acceleration

If GPU acceleration isn’t available, EasyWhisperUI can still run using CPU only (slow).

---

Custom model support

EasyWhisperUI supports custom Whisper models now!

Use a custom model

1. Open the Model selector and select custom
2. Choose Select Model File
3. Pick your local model file (.gguf/.ggml/.bin)
4. Start transcribing

---

Requirements

Windows 10/11

• AMD / Intel / NVIDIA GPU with Vulkan support
  (Most modern discrete + integrated GPUs work.)
• Virtual machines require Vulkan support (e.g., GPU passthrough)

macOS

• Apple Silicon (M1 / M2 / M3 / M4 / M5)

Linux

• Linux support is available
• Please share feedback/issues if something breaks on your distro or hardware

---

Install

Windows

1. Download the latest Windows installer from Releases
2. Run it (installs per-user under app data)
3. Desktop + Start Menu shortcuts are created

macOS

1. Download the .dmg from Releases
2. Open it and drag EasyWhisperUI into Applications

Linux

1. Download one of the Linux artifacts from Releases (.AppImage, .deb, or .rpm)
2. Install/run it using your distro’s normal workflow
3. On first launch, EasyWhisperUI installs required dependencies and compiles Whisper binaries
4. If anything fails, please open an issue with logs

---

Donate

Thanks to supporters ❤️

• Craig H: $50
• Jan P: $5
• Minh P: $5
• Rödvarg R: $2

This project takes a lot of time to maintain and test across systems. If EasyWhisperUI helped you, consider supporting development:

👉 Donate via PayPal
https://www.paypal.com/donate/?business=5FM6Y27A3CK58&no_recurring=0&currency_code=USD

---

Credits

• whisper.cpp by Georgi Gerganov
• FFmpeg
• Windows FFmpeg builds: gyan.dev
• Packaging: electron-builder

---

License

Copyright (c) 2025 Mehtab Mahir
All rights reserved.

This software is proprietary and the following is not allowed for commercial purposes:
it may not be copied, modified, distributed, or used without explicit permission from the author.

Those actions are permitted for personal use ONLY.

This application includes the following open-source components:

---

whisper.cpp by Georgi Gerganov
License: MIT
[https://github.com/ggerganov/whisper.cpp](https://github.com/ggerganov/whisper.cpp)

---

FFmpeg
License: LGPL 2.1
[https://ffmpeg.org](https://ffmpeg.org)
Windows builds by: [https://www.gyan.dev/ffmpeg/](https://www.gyan.dev/ffmpeg/)

The FFmpeg binary is provided as a separate file and may be replaced with a compatible version.

---

Build steps (for developers)

Electron app (TypeScript)

From the electron/ folder:

npm install

Package (generates installer artifacts):

```bash
npm run dist

Build output:

• build/electron-dist