Open Wispr Flow — Cloud Edition

AI-powered voice dictation for Windows & macOS — speak naturally and get clean, context-aware text pasted at your cursor. An open source wisprflow.ai alternative

How It Works

Hold Ctrl+CMD(Win) → Speak → Release → Text appears at cursor

1. Record — Captures audio while you hold the hotkey
2. Transcribe — Sends audio to OpenAI Whisper API
3. Detect Context — Reads the title of your active window (VS Code, Chrome, Slack, etc.)
4. Correct — GPT fixes transcription errors using the window context
5. Paste — Clean text is pasted at your cursor position

You can install the executables from the dist folder. Set the api key and press ctrl followed by CMD9Win) key and start speaking. You can edit settings in the tray application and update the promt as you see fit.

Quick Start

Prerequisites

• Python 3.9+ — Download
• OpenAI API Key — Get one here

---

Windows

cd open_wispr_flow\cloud
pip install -r requirements.txt
$env:OPENAI_API_KEY = "sk-your-key-here"
python main.py

macOS

cd open_wispr_flow/cloud
pip3 install -r requirements_mac.txt
export OPENAI_API_KEY=sk-your-key-here
python3 main.py

⚠️ macOS Permissions Required: Go to System Settings → Privacy & Security and grant:

• Accessibility — for global hotkey listener
• Microphone — for audio recording
• Input Monitoring — for keyboard events

---

Interactive Mode (for testing)

python main.py --interactive

Building a Standalone App

Windows → EXE (standalone)

build_exe.bat

macOS → Binary (standalone)

chmod +x build_mac.sh && ./build_mac.sh

---

Creating Installers

Windows Installer (.exe setup wizard)

Creates a proper installer with Start Menu shortcut, Desktop icon, optional auto-start, and API key prompt during install.

Prerequisites: Install Inno Setup 6 (free)

REM One-click build (builds EXE + installer)
build_installer_win.bat

Or step by step:

1. Build the EXE: build_exe.bat
2. Open installer_win.iss in Inno Setup Compiler
3. Click Build → Compile
4. Installer at: installer_output\OpenWisprFlow_Setup.exe

Launch sequence is ctrl+Win. Hold and speak and release to get transcription in the open textarea

macOS Installer (.dmg with drag-to-Applications)

Creates a .app bundle with native API key dialog on first launch, packaged in a .dmg.

chmod +x build_installer_mac.sh
./build_installer_mac.sh

Output: dist/OpenWisprFlow_Installer.dmg

User installs by:

1. Open the .dmg
2. Drag OpenWisprFlow to Applications
3. Make sure you give input tools, microphone and accessibility permissions in privacy settings after installing.
4. Launch from Applications
5. Enter API key when prompted (saved for future runs)

Launch sequence is ctrl+cmd. Hold and speak and release to get transcription in the open textarea

Configuration

Settings are stored at ~/.open_wispr_flow/config.json. You can also edit config.py directly before building.

Setting	Default	Options
GPT_MODEL	gpt-4o-mini	gpt-4o, gpt-4o-mini, gpt-3.5-turbo
WHISPER_MODEL	whisper-1	whisper-1
RECORD_MODE	push_to_talk	push_to_talk, toggle
HOTKEY	Ctrl+Win	Configurable in config.py

Project Structure

cloud/
├── main.py              # Entry point + hotkey listener
├── config.py            # Configuration
├── recorder.py          # Audio recording (sounddevice)
├── transcriber.py       # OpenAI Whisper API
├── window_context.py    # Active window detection (cross-platform)
├── llm_corrector.py     # OpenAI GPT correction
├── text_paster.py       # Clipboard + paste simulation
├── build_exe.bat        # Windows EXE build script
├── build_exe.py         # Python EXE build script
├── build_mac.sh         # macOS build script
├── requirements.txt     # Windows dependencies
├── requirements_mac.txt # macOS dependencies
└── README.md            # This file

Costs

Component	Cost
Whisper (30s audio)	~$0.006
GPT-4o-mini correction	~$0.0003
Total per dictation	~$0.007

That's roughly 140 dictations per $1.

Troubleshooting

Issue	Platform	Fix
OPENAI_API_KEY not set	Both	Set the environment variable before running
No audio captured	Windows	Check microphone in Sound Settings
No audio captured	macOS	Grant Microphone permission in System Settings
win32gui not found	Windows	pip install pywin32
Hotkey not working	macOS	Grant Accessibility & Input Monitoring permissions
EXE/binary won't start	Both	Run from terminal to see error messages
Paste not working	Both	Make sure a text field has focus before releasing hotkey