Whisper Typer

Push-to-talk voice transcription using Faster-Whisper. Supports Windows, macOS, and Linux.

Quick Start

1. Start the app:

   uv run run.py

2. In the app:

   • The server auto-starts on launch.
   • Choose a Model and Input Mode (Live or Full Capture).
   • Press Cmd+G (macOS) to start/stop recording.
   • On Windows, Win+G may be intercepted by Xbox Game Bar; if so, use the in-app Start Recording button or disable the Game Bar shortcut in Windows settings.
   • Text types into your active window automatically.

---

Installation

If you want to install it as a global tool:

uv pip install -e .
whisper-typer

---

Flow logic

%%{init: {"flowchart": {"htmlLabels": false}} }%%
flowchart TD
    A["User Hotkey"] --> B["Audio Input Stream"]
    C{"Input Mode"}
    C -->|Live typing| D["Silence-based Chunking"]
    C -->|Full Capture| E["Full Recording Capture"]
    D --> F["Transcription Queue (FIFO)"]

    E --> F
    F --> G["Server API (Transcribe)"]
    G --> H["Transcription Service"]
    H --> I["Text Output"]
    I --> J["Keyboard Typing to Active Window"]

Loading

• User presses Win+G to toggle recording.
• Audio is captured from input stream.
• App checks selected mode:
  • Live typing → chunks split by silence windows and enqueued.
  • Full Capture → all chunks captured until stop, then enqueued.
• Queue processes each chunk in order (FIFO).
• For each chunk:
  • Send audio to server via API.
  • Server returns transcribed text.
  • Text is typed into the active window via keyboard simulation.

---

Hotkeys & Auto-typing

The client runs a global hotkey listener:

• Cmd+G (macOS) — Toggle recording.
• Win+G (Windows) may be reserved by Xbox Game Bar; if it does not work, use the in-app Start Recording button.
• When recording is stopped, the client waits for the transcription and then simulates keyboard typing to insert the text into the currently focused window.

macOS Users:

1. You must grant Accessibility permissions to your terminal (e.g., iTerm or Terminal.app) for the auto-typing to work.
2. Grant Microphone permissions when prompted.

System tray icon colors

State	Color	Meaning
Idle (server online)	🟢 Green	Server is running, ready to transcribe
Server offline	⚫ Black	Server is not reachable
Recording	🔴 Red	Audio is being captured
Processing	🟣 Purple	Transcribing audio

---

Requirements

• OS: Windows, macOS, or Linux
• Python: 3.10+
• Package manager: uv (recommended)
• Docker: Optional, for isolated container deployment

---

Configuration

The application stores data in ~/.whisper-typer/ by default. You can customize settings using a .env file in the project root:

• WHISPER_MODEL: Default model (e.g., tiny, small, medium).
• WHISPER_MODELS_DIR: Custom path for model storage. Use an absolute path (for example D:/AI/whisper-models on Windows or /absolute/path/to/models on Linux/macOS) so the client and server always use the same directory.
• HF_TOKEN: Hugging Face token for private models.

---

Contributing

Contributions are welcome! Please see CONTRIBUTING.md for details.

---

License

This project is licensed under the Apache License 2.0. See the LICENSE file for details.

---

About the Author

Sharad Raj Singh Maurya
AI Engineer and Open Source enthusiast.

• GitHub: @sharadcodes
• Project: Whisper Typer

Feel free to reach out for collaborations or to report any issues!